Bio-based and Applied Economics 8(2): 179-210, 2019

ISSN 2280-6180 (print) © Firenze University Press 
ISSN 2280-6172 (online) www.fupress.com/bae

Full Research Article

DOI: 10.13128/bae-8930

Innovation adoption and farm profitability: what role for 
research and information sources?

Michele Vollaro1,*, Meri raggi2, DaViDe Viaggi1

1 University of Bologna/Department of Agricultural Sciences, Agricultural Economics, Bologna, Italy 
2 University of Bologna/Department of Statistical Sciences, Bologna, Italy

Abstract. The paper analyses the determinants of farmers’ adoption of innovations 
and studies the effect of the source of information and the connection with agricul-
tural research on the contribution of innovation to farm performance. The paper uses 
primary data collected ad hoc in the Province of Bologna (Emilia-Romagna, Italy) and 
analyses it through an econometric analysis. The results indicate that structural factors 
and farm specialisation still play a relevant role in innovation adoption. Connection to 
scientific research triggers significant improvements in terms of value-added and qual-
ity of production but does not affect other profitability-related parameters. The results 
confirm the need for policy to better consider the role of intermediate actors between 
research and the farmer as well as to better clarify the final performance strategy in 
order to set the policy instruments right. The paper also highlights the need for fur-
ther research about farms’ proactivity in searching for and selecting information dur-
ing the process of innovation adoption and competitive advantages in terms of profit-
ability components.

Keywords. Innovation adoption, information sources, research-innovation link, 
farm profitability.

JEL Codes. D83, O14, O31, O33.

1. Introduction

The interest in studying the process of innovation adoption and impact, both from 
theoretical and empirical perspectives, is motivated by the key role of innovation in fos-
tering agricultural competitiveness and socio-economic growth (Ramos-Sandoval et al., 
2018; Sauer et al., 2019). In fact, a noteworthy share of the literature to date has focused 
primarily on understanding the patterns of innovation diffusion, rather than adoption. In 
recent decades, several studies have started to broaden the research perspectives on agri-
cultural innovation by introducing frameworks and models aimed at understanding the 
process of innovation adoption in agriculture (Gadhim and Pannell, 1999; Diederen et al., 
2002). The early approaches can be roughly classified into those mostly focusing on eco-

*Corresponding author: michele.vollaro@unibo.it



180 Michele Vollaro, Meri Raggi, Davide Viaggi

nomic interpretations and those taking a more sociological perspective (Marra et al., 2003). 
Economists have argued that adoption and diffusion of innovation is motivated by changes 
in economic factors, such as prices, production efficiency, risk attitude and utility, whilst 
sociologists have, for their part, highlighted the major role of the adopter’s characteristics 
and the social environment in which the adoption process occurs. Although from different 
perspectives, both approaches have converged in identifying in the learning behaviour of 
individuals one of the most important factors in the innovation adoption process which, in 
turn, characterises the diffusion pattern (rate of adoption) (Ruttan, 1996). 

Micro-level studies concerning the adoption and diffusion of innovations has pro-
gressed over time by testing new models explaining adoption and new patterns of dif-
fusion, characterised by the inclusion of (farm level) information (uncertainty and 
risk) and time (diffusion as a sequence of adoptions) factors. While new insights have 
been identified with respect to the theoretical evolution, the empirical results provide 
an increasingly varied range of explanations for adoption (Ghadim and Pannell, 1999), 
including the recent attention to the innovation behaviour of farmers (Läpple et al., 2015; 
Sauer et al., 2019).

At the aggregate level, the evolution of theories and practices concerning the concept 
of innovation have moved from a linear model of knowledge transfer from public research 
to the farm (Röling, 1990), to the so-called agricultural knowledge system (AKS), to an 
even more complex and dynamic innovation process, in which different actors (including 
public and private stakeholders beyond the research, education and consulting/extension 
sectors) cooperate in a network, referred to as agricultural knowledge and (information) 
innovation system (AKIS) (Esposti, 2012; SCAR, 2012; Ramos-Sandoval et al, 2019). The 
AKIS concept supports the idea that the development and realisation of innovations are 
not limited to pre-defined and unidirectional processes (path-dependence, demand-pull 
or technology-push), as in the case of AKS, but rather fed by a multitude of processes 
characterised by a continuous interaction between stakeholders within a network. Such a 
paradigm although, on the one hand, makes the study of approaches to innovation adop-
tion more complex, on the other hand it broadens the research perspectives by allowing 
for the inclusion of latent or hidden elements in modelling innovation adoption in agri-
culture, such as multiple information channels, for which the contribution of literature is 
still limited, and the role of research, recently highlighted by several European policy ini-
tiatives such as the European Innovation Partnership (EIP) and the (related) Innovation 
Operation Groups (IOG). An especially relevant gap in the literature concerns the link 
between upstream connections with research as a source of information and innovation 
performance on the farm, in a context characterised by the growing role of the farmer in 
combining information and new technologies in designing farm-level innovations.

This paper seeks to contribute to this literature through a farm-level study on the 
impacts of scientific research in agriculture (SRA) on the economic performance of farms 
taking into account the intermediary steps of innovation adoption. The paper relies main-
ly on primary farm-level data, collected through direct interviews with farmers using an 
ad hoc survey questionnaire, with the broad aim of collecting data suitable for analysing 
the determinants of farmers’ adoption of innovations, the effect of innovation on farm 
economic outcomes, in terms of different components of profitability of the introduced 
innovation, and link these back to the role of research in innovation development.



181Innovation adoption and farm profitability: what role for research and information sources?

The contribution of the paper is mainly on empirical grounds, using insights and vari-
ables from a wide range of literature. However, our study is inspired by concepts mainly 
derived from two seminal theoretical frameworks, namely induced technical change by 
Hayami and Ruttan (1985) and the evolutionary model by Nelson and Winter (1982). We 
apply a demand-driven approach, as proposed by Walker et al. (2010), which, together 
with the recall technique strategy, allows us to set an impact pathway, tracing back the 
determinants of the effects of successful innovation adoption on economic performance. 
In this paper, the use of the term innovation is intended as new to the farm/farmer and not 
as new to the market (Mairesse and Mohnen, 2010).

The main novelty of the paper is the attempt to clarify whether a higher farm per-
formance might be linked to the fact that the adopted innovation is rooted in scientific 
research. In particular, we investigate the extent to which, and how, the fact that an inno-
vation is known to derive from scientific research affects, beyond the adoption decision, 
the economic performance of the farm. The origin of innovation from scientific research 
is identified through collected data about prior-knowledge of farmers. Performance is also 
measured based on farmers’ statements regarding the gains realised from the adoption of 
the innovation, in terms of reduced costs, increased production, higher value-added and 
higher product quality. 

We investigate the effects of information on the adoption decision processes in two 
steps. After having presented the survey results in terms of descriptive statistics, we first 
analyse which factors and processes influence farmers’ decisions to adopt, or not to adopt, 
new technologies; then, as concerns the innovators (the farmers who introduced an inno-
vation), we investigate whether the origin of innovation from scientific research yielded 
effects on profitability at farm level. 

The paper continues with a literature review in section 2. The methodology is out-
lined in section 3, followed by the presentation of the case study area (Province of Bolo-
gna, Emilia-Romagna) in section 4. Section 5 illustrates the results, followed by a discus-
sion in section 6 and concluding remarks in section 7.

2. Literature review

Early studies on innovation adoption at farm level focused mostly on disentangling 
the innovation adoption process through a micro-economic approach (Cochrane, 1958; 
Hayami and Ruttan, 1985; Thirtle, 1985), by relying on the basic assumption of profit 
maximisation as the main economic driver for adoption (Sunding and Zilberman, 2001). 
On the other side, recent studies on innovation in agriculture, although relying on the 
same framework, focus more on the variety of different elements determining the adop-
tion, as well as the diffusion processes. Indeed, Hall (2012) sketches how the modern 
innovation adoption process goes largely beyond the (public) function of introducing 
technology to farmers, as exogenously intended by Cochrane (1958) and Hayami and Rut-
tan (1985), conceiving the innovation in agriculture as a system in which partnership, alli-
ance and network actors work together to develop and spread innovation. A fundamental 
role in this system is acknowledged to be played by producers and users of knowledge, but 
the issue of who and how such links are created is still very much under scrutiny. Indeed, 
the multiplicity of underlying dynamics characterising the links between the actors of the 



182 Michele Vollaro, Meri Raggi, Davide Viaggi

agricultural innovation system might be at the basis of the discordant findings of stud-
ies on innovation adoption. In fact, recent studies on the topic have addressed this issue 
and are evolving towards the definition and role of knowledge and/or innovation brokers 
within the AKIS (Klerkx et al., 2009; Ramos-Sandoval et al, 2019). 

As regards the adoption of innovation in agriculture, in fact, different studies report 
varied results with regard to the relative importance of different determinants of adop-
tion (Ghadim and Pannell, 1999), such as education, credit constraints, land size and oth-
ers (Feder and Umali, 1993). One reason for such discordant results can be attributed to 
the difficulty in relating empirical information, model hypotheses and the conceptual/
theoretical framework in which innovation adoption in agriculture is modelled (Lindner, 
1987; Besley and Case, 1993). In fact, the evolution of the theoretical framework pro-
gressed towards the inclusion of informational attributes (Koundouri et al., 2006; Walder 
et al., 2019) and learning behaviour (Ramos-Sandoval et al., 2018) into the models hence 
making it possible to envisage innovation adoption as a dynamic process. Information has 
played a major role in modelling the uncertainty concerning adoption decisions as well as 
farmers’ risk attitudes and risk aversion behaviour in the face of uncertainty. Indeed, in a 
context of incomplete information, the degree of risk perception is assumed to be affected 
by learning, as learning can reduce the uncertainty concerning the innovation adoption 
(especially the downside production risk) (Marra et al., 2003; Koundouri et al., 2006). 
Time, especially in connection to learning, is another important factor characterising 
the speed and rate of aggregate adoption and, hence, diffusion (Sunding and Zilberman, 
2001). Other additional factors beyond profitability, such as environmental and social sus-
tainability concerns, potentially determining innovation adoption have been explored as 
well (Walder et al., 2019). 

In relation to the above, diffusion itself has been subject to different interpretations. In 
economic terms it can still be interpreted as depending mostly on the perceived short-run 
profitability of the innovation (Levins and Cochrane, 1996; Diederen et al., 2002). How-
ever, from a more sociological perspective, innovation diffusion also depends on the spread 
of information and is negatively related to the distance from the propagation point (Rog-
ers, 1983). Improvements in human capital through learning affect positively the adoption 
rate and diffusion of innovation. Based on this concept and starting from the evolutionary 
model of Nelson and Winter (1982), a stream of research advanced up to the adaptation of 
the technology acceptance model (TAM), proposed by Davis (1989), to the farming sector 
(Flett et al., 2004; Folorunso et al., 2008; Rezaei-Moghaddam and Salehi, 2010). Through 
TAM, innovation adoption is explained as a process that depends on the perceived useful-
ness and perceived ease of use of the technology which, in turn, affects the acceptance (and 
the adoption) of the innovation. This theoretical framework belongs mainly to the psycho-
logical perspectives of the topic and attributes more importance to the individual beliefs 
and perceptions underlying the learning behaviour involved in the adoption process.

A noteworthy gap in the literature concerns investigating whether and how the origin 
of innovation, and in particular research, may be related to the economic performance of 
innovation adopted by the farms1. Two aspects can be distinguished: a) one is the “objec-

1 This topic has been recently explored by Hockmann et al. (2018) in the food processing sector, evaluating the 
impact of internal R&D activities on the economic performance of multinational corporations.



183Innovation adoption and farm profitability: what role for research and information sources?

tive” origin of innovation; and b) the second is the knowledge of the origin. This distinc-
tion and the attribution of innovation to specific events or projects is often difficult due to 
the fact that multiple players and activities may contribute to its development, including 
the farmers themselves.

As demonstrated in the literature, knowledge about innovation, improved through 
learning, plays a central role in the adoption process (Marra et al., 2003). This holds espe-
cially in agriculture where the relatively high costs of internal R&D activities do not allow 
for the easy and affordable development of innovations within the farm (Sunding and Zil-
berman, 2001; Diederen et al., 2003). For a large part of the literature, the positive role of 
knowledge by the farmer in the process of innovation adoption is referred to (or limited 
to) the adoption of available innovations and limited to the features of innovative technol-
ogies/solutions, disregarding its origin. In particular, learning (ability) is mostly consid-
ered to be a skill that makes the farmer (the innovator) able to reduce the downside risk 
of the innovation adoption and to improve the performance of the innovation through a 
process of adaptation to his/her farm’s peculiar characteristics.

This approach implicitly assumes that the learning behaviour is considered to be 
detached from the path leading from research to innovation. Instead, here we assume that 
knowledge about the innovation development process matters in terms of improved adop-
tion processes and economic performance of the farm. This view is consistent with theo-
retical frameworks and empirical evidence that highlight how the cognitive elements of the 
innovator, such as his/her educational background (Lin, 1991; Reimers and Klasen, 2013), 
attainments and experience (Foster and Rosenzweig, 1995), affect positively both adoption 
and performance of  innovations (Sauer et al., 2019), though (knowledge about) origin is not 
generally explicitly addressed. Moreover, this hypothesis easily accommodates the theoretical 
framework pertaining to AKIS, according to which farmers interact with articulated networks 
of actors in the innovation research, development and adoption processes and may hence be 
aware of, or participate in, the research stages of innovation development or in the further 
stages of knowledge dissemination (SCAR, 2012). In this paper, we consider both the knowl-
edge of research generating the innovation and the sources of information about the innova-
tion as potential factors affecting the economic performance of innovation adoption.

3. Methodology

3.1 Overall approach 

The analysis proposed in this paper seeks to link information, research and farm-lev-
el performance by analysing the declared effects of innovation introduction with respect 
to farm structural factors, farmers’ characteristics and elements related to the process 
of innovation adoption, such as the sources of information, specifically the origin from 
research. The main objective of our study (besides explaining innovation adoption) is to 
evaluate whether the effects of adopted innovation on farm profitability is affected by the 
origin of the adopted innovation, in particular how innovation originating from research 
can affect various aspects of farm performance in different ways.

The paper is based on survey data, provided from farmers’ responses to questions. 
This will require some qualifications, which are provided in the discussion section. 



184 Michele Vollaro, Meri Raggi, Davide Viaggi

As mentioned, the paper does not refer to one specific theoretical framework. However, 
the set of explanatory variables draws mainly from the analysed literature, which is ground-
ed upon the  induced technical change theory by Hayami and Ruttan (1985), for which inno-
vation adoption is responsive to both economic conjuncture and technical evolution brought 
about by R&D, and the evolutionary model (Nelson and Winter, 1982), according to which 
farmers put effort into searching for better techniques and the selection of successful inno-
vations (local searches for innovations, imitation of the practices of others and satisficing 
economic behaviour). These theoretical frameworks are integrated with insights drawn from 
the most recent literature on the AKIS framework and innovation adoption, especially con-
sidering linkages with non-farm actors, different sources of information and personal atti-
tudes towards adoption. The theoretical development of the topic involves further aspects of 
the process in order to better qualify innovation adoption and diffusion, such as diffusion 
in terms of imitation of adoption, timing of adoption, endogenous and exogenous factors 
affecting adoption, elements characterising heterogeneity of farmers, etc.

In order to address the evolving theoretical framework and to adapt to available data, 
a variety of methodological approaches have been used in the literature. Sunding and 
Zilbermann (2001), in reviewing the innovation process in agriculture, argued that the 
analytical methodologies mostly suited to evaluate the process of technology/innovation 
adoption are the binary or the limited dependent variable approaches. This opinion hinges 
upon the fact that innovation adoption is regarded as a discrete choice and, as such, rep-
resented by the means of threshold models. Alternative and more articulated approaches 
have been employed over time, e.g. Ghadim and Pannell (1999) adopted time-series meth-
odologies, Diederen et al. (2002, 2003) used nested and ordered logit models, Dimara and 
Skuras (2003) tested the application of partial observability models, while Koundouri et al 
(2006) applied a two-stage binary choice model. 

Given that the objectives of the present study mainly pertain to the evaluation of 
the effects of different elements of the innovation adoption process on both the adoption 
choice itself and the consequences of the adoption in terms of positive economic perfor-
mance, we use econometric techniques belonging to the class of limited dependent vari-
able models on cross-section data derived from an original survey.

3.2 Methodological approach 

A two-stage conceptual framework is employed for modelling the analysis. The first 
stage concerns farmers’ choice to adopt an innovation and the second concerns the profit-
ability of the adopted innovation. The underlying process is composed by a participation 
stage and an outcome stage, where the outcome depends on the participation: the first 
stage is about the choice to adopt or not and, conditional on this first decision, the second 
is about the economic performance of the adopted innovation. 

An expected utility maximization framework is used to examine farmers’ choice to 
adopt, including the sequential adoptions as well. Assuming that farmers are profit ori-
ented and that their expected utility depends on the level of profit earned, the objective 
function of the farmers will be to maximize expected utility through maximizing expected 
profits (posed that utility is monotonically increasing in expected profit). It follows that a 
higher profit implies a higher expected utility for farmers.



185Innovation adoption and farm profitability: what role for research and information sources?

Thus, for the ith farmer: Ui=U(πi(Ii,Xi,Si)), where Ui is expected utility of farmer i, πi 
is expected profit of farmer i, Ii is the innovation adopted (that guaranteed the highest 
performance) by farmer i, Xi is vector of determinants of adoption of farmer i that impact 
expected profits of production, and Si is a vector of other factors affecting the ability of 
farmer i of generating profit.

According to Lynes et al. (2016), the choice of adopting an innovation occurs if the 
expected utility Ui, expressed in terms of expected profit from the adoption of Ii, is great-
er than the expected utility of no adoption, namely no Ii. Assuming that the choice of Ii 
depends on Xi and Si as well, Ii(Xi,Si) and by simplifying the notation, so that Ui is stated 
as a function of Ii, the following condition applies: Ui(Ii)>Ui(no Ii), such that Ui(Ii)–Ui(no 
Ii)= ∆(Ui)>0. 

Expected higher profits, i.e. the outcome, is dependent on the choice of adopting, i.e. 
the participation. The outcome stage can be identified according to two different specifica-
tions. On one hand, the outcome of adopting an innovation, as suggested by Cochrane 
(1958) and Levins et al. (1996), can be intended as a continuous choice or a sequence of 
adoptions, namely more than one adoption, in order to guarantee, according to the tech-
nology treadmill, the competitiveness and the profitability of the farm. On the other hand, 
the outcome stage can be meant as the profitability consequent to the adoption of a specif-
ic innovation, namely the realized economic performance resulting from the introduction 
of the innovation into the farm. 

In both cases, it is assumed that farmers who choose to adopt knows that the out-
come is affected by adoption determinants, such as structural factor (farm size, specializa-
tion, mechanization, market), subjective characteristics of the farmer (education, experi-
ence, off-farm income, business motivation, entrepreneurial attitude), but they also know 
that, to maximize the profitability, innovations need to be introduced after a learning 
process has been made and after that other elements have been scrutinized and evaluated 
accurately, such as the ability of self-developing the innovation, trial and error, the sources 
of information from others and links with R&D. Expected higher profit can, therefore, be 
considered as an indirect function of both the determinants, the farmers’ subjective char-
acteristics and the learning process leading to the adoption of a specific innovation. The 
stage two can be represented as follows: πi[Ii(Xi,Si)]>0, for which >0, >0 and, in turn, 

>0, while  is ambiguous. 

3.3 Econometric modelling strategy 

The econometric modelling strategy proceeds in two main steps: first, we provide an 
analysis of the adoption choice; then we proceed by explaining the performance and con-
necting it to the source of information. In order to avoid potential confusion across analy-
ses and models, the first group is called adoption models, while the second is referred to as 
performance models.

The analytical models chosen to analyse such variables belong to the class of limited 
dependent variable models. In the general case, the choice to adopt, namely the adoption 
model, is observed as a binary action, representing the underlying outcome of the utility 
maximization: if Yai=1 means that ∆(Ui)>0, while in the opposite case Yai=0. That is, Yai=1 
when farmer i chooses to adopt the innovation, and Yai=0 otherwise. Determinants of (Xi) 



186 Michele Vollaro, Meri Raggi, Davide Viaggi

and other factors (Si) are assumed to linearly affect the adoption decision related to the 
farmers’ choice to adopt. Let Zai(Za1,…,Zak) be the set of both the determinants of (Xi) and 
the other factors (Si) affecting the adoption choice, aai(aa1,…,aak) be a vector of parameters 
and εi be a mean zero IID error term. 

Then, the adoption choice can be modelled as: .
The choice variable is simply the record of the adoptions, recorded as a single choice 

(in the case of one innovation) and as a sequence of choices (in the case of more than one 
innovation). This part of the analysis was carried out by evaluating the determinants of 
both the propensity to innovate and the number of innovations introduced, by employing 
a Probit and a Poisson model, respectively. In addition, a Double-hurdle model has been 
used. This type of model has the advantage of making it possible to analyse the number of 
adoptions (single or repeated) that are conditional on the analysis of the choice to inno-
vate (participation), which potentially follows a different data generating process (or, rath-
er, that may be affected by different explanatory variables). The additional contribution of 
the double-hurdle regression is the capacity to clearly separate the factors mainly affecting 
the choice from those mostly affecting the adoption. The determinants include the tech-
nical and commercial characteristics of the farms and the subjective, socio-demographic 
characteristics of the farmers. Other factors include the motivations of farmers to inno-
vate, the knowledge of the adopted innovation prior to its adoption, the sources of infor-
mation that farmers consulted, including the origin of innovation from scientific research, 
as well as whether farmers developed the innovation by themselves. 

Following the same rationale, the profitability induced by the adopted innovation, 
namely the performance model, is observed as a binary outcome as well: if Ybi=1 means 
that >0, while in the opposite case Ybi=0. That is, Ybi=1 when the adopted innovation 
yielded an improvement in profitability and Ybi=0 otherwise. Even in this case, determi-
nants of (Xi) and other factors (Si) are assumed to linearly affect the improvement in prof-
itability. Let Zbi(Zb1,…,Zbk) be the set of both the determinants of (Xi) and the other fac-
tors (Si) affecting the profitability (they do not need to be the same employed in step one), 
αbi(αb1,…,αbk) be a vector of parameters and ξi be a mean zero IID error term. 

Then, the profitability can be modelled as: .
The determinants are the same as the previous analytical model, while other factors 

include the motivations of farmers to innovate, the knowledge of the adopted innovation 
prior to its adoption, the sources of information that farmers consulted, including the ori-
gin of innovation from scientific research, as well as whether farmers developed the inno-
vation by themselves. 

Profitability is the measure of the realized gains, based on farmers’ declarations, 
resulting from the introduction of the innovation, in terms of cost reduction, production 
increase, value-added increase and quality increase. The first three have been collected in 
per cent terms, while the last in ordinal categorical terms (not at all, low, high, very high). 
However, they have been transformed in binary variables in order to evaluate solely the 
presence (not the magnitude) of the declared (positive) effects of the introduced inno-
vation. However, given the use of the recall technique, these variables could suffer from 
approximation due to difficulties in providing precise estimates of the actual amount 
(Mairesse and Mohnen, 2010). Such potential measurement errors could lead to biases in 
estimates and inefficient statistical conclusions and, in turn, render the use of the Tobit 



187Innovation adoption and farm profitability: what role for research and information sources?

model ineffective. Despite this, these data provide for (i) important quantitative informa-
tion, when used for explorative descriptive statistics and for comparative exercises, and (ii) 
qualitative information, when opportunely transformed into binary or categorical vari-
ables, to be used in econometric models for inferential purposes. Indeed, the hypothesis 
of experiencing better performance if the innovator knows that the adopted innovation is 
derived from research could be reformulated in terms of positive (or non-null) performanc-
es. This implies the cost of losing the magnitude of the effect (marginal effect) but, at the 
same time, the benefit of at least keeping the presence of the effect (propensity of experi-
encing a positive outcome). 

Such a perspective makes it possible to approach the analysis by considering the 
measured performance in terms of latent continuous variables and, in turn, by employ-
ing a Probit and a Heckit model, with the aim, respectively, of analysing the propensity of 
obtaining positive performances, with regard to the innovators, and of accounting for the 
possible presence of sample selection bias. In fact, the presence of positive performance 
outcomes due to the research-innovation link might depend upon the self-selection pro-
cess of those farmers who decided to innovate because of higher expected gains. Each 
model has been applied separately to each of the four performance variables, using the 
same set of explanatory variables.

The analysis on economic performance has the same specification of the probit adop-
tion regression with the inclusion of the other factors, namely the variables accounting for 
knowledge of the research-innovation link and source of information (hereafter “informa-
tion variables”). Specifically, the research-innovation link is the variable expressing whether 
the farmer is informed that the innovation originated from research, while source of infor-
mation indicates whether the farmer knew about the innovation from external sources or 
developed the innovation by himself. Age of innovation, for its part, is a measure of time 
distance between the year of introduction and 2015 (maximum 20 years) and is a proxy of 
farmers’ experience using such innovation (fine-tuning of innovation usage) as well as for 
the innovation to fully express its effects in terms of economic performance. The depend-
ent variables used in the probit performance models are cost reduction, production increase, 
value-added increase and quality increment, all expressed as binary variables.

3.4 Survey: sampling procedure and questionnaire 

A survey strategy was adopted because of the absence of datasets on innovation adop-
tion processes and/or the existence of datasets characterised by noteworthy margins of 
non-representativeness and of collection/transcription errors, such as the ones operated by 
regional administrations to evaluate measures of the Rural Development Plans (RDP) or 
the regional level FADN data. The survey strategy represents an appropriate research tool 
for this work because, like similar research works, this type of approach is preferred for 
anticipatory/forecast purposes and for studying elements and factors that are much more 
difficult to identify, such as the innovation adoption process (Besley and Case, 1993).

The sampling plan, aimed at collecting complete information from at least 300 farms 
in the Province of Bologna, randomly picked from a sequential selection of about 1000 
farms, constrained to be representative of both the agricultural specialization (type of 
farming) and the altitude level. 



188 Michele Vollaro, Meri Raggi, Davide Viaggi

The data have been collected by the way of an ad hoc questionnaire, first checked 
through direct interviews, further adapted to be used by telephone and finally carried out 
by telephone interviews (of approximately 15 minutes in length). 

The survey was designed to collect information about the farm, information about 
the farmer, specific elements pertaining to the innovation adoption process realised by 
the farmer and, in sequence, the relative effects on farms’ economic performance from the 
adopted innovation. 

The questionnaire is structured in six sections:
• The introduction presents the aims of the survey and the project it relates to (EU FP7 

project IMPRESA);
• The first section includes questions about farm structure: production specialisation 

and ancillary activities; land, labour, machines, technological plants;
• The second section deals with the adoption process, including the choice of innovat-

ing, the number and types of innovations introduced, the motivation for, and for not, 
innovating;

• The third section concerns one introduced innovation, namely the most important 
innovation (in terms of profitability), the sources of information and the link with 
research;

• The fourth section addresses the financial aspects of innovation adoption, in particu-
lar whether the innovators benefited from supports from the Common Agricultural 
Policy (CAP) and the amount of total investments;

• The fifth section deals with the effects of the adopted innovation in terms of econom-
ic performance: perceived changes in costs (efficiency gains), in production (output 
gains), in value-added products and (higher) product quality; 

• The sixth section includes questions about future behaviour of the farmers and expec-
tations/sentiments with respect to the CAP;

• The last seventh and final section includes questions about the socio-demographic 
characteristics of the farm and the farmer’s family.
The first and the last sections of the questionnaire aim to collect, respectively, struc-

tural (objective) data about the farms and socio-demographic (subjective) data about the 
farmers, focusing on those elements considered in the literature as “classic” determinants 
of innovation adoption, such as specialisation, size, mechanization, altitude, farm income, 
education, and experience. The second section inquiries into the process of innovation 
adoption by first exploring (eliciting) the opinion of the farmer about the existence of 
important innovations (in terms of profitability) in his specialisation sector in the last 20 
years. The subsequent information regards the types of innovation introduced on the farm 
in the last 20 years, as well as the choice of not introducing any particular innovation, 
specifying innovation with regard to products, production factors and process innova-
tions. Crossing these two types of information makes it possible to clearly frame the indi-
vidual choice context in which the adoption process has been developed. In this section, 
the farmer indicates which of the introduced innovations is, in his/her view, the most 
important in terms of profitability. The third section focuses solely on the most impor-
tant innovation indicated by the farmer and deals mainly with the motivations underlying 
the adoption. This section has been built on the basis of the Induced Innovation Adoption 
(IIA) by Hayami and Ruttan (1985) and the evolutionary model (EM) by Nelson and Win-



189Innovation adoption and farm profitability: what role for research and information sources?

ter (1982). With regard to the IIA, farmers were asked whether the choice of innovating 
was determined, inter alia, also by a reaction to changes in products’ and factors’ prices or 
by the intention to anticipate the evolution of the markets of both products and produc-
tion factors. Further, the condition of being early adopters or laggards has been investigat-
ed by asking farmers for how long the introduced innovation was already commonly used. 
As regards the EM, farmers were questioned about the origin of the introduced innova-
tions, with the aim of exploring the connections between the farmer and the other actors 
involved in the AKIS, including the research sector. In primis, a distinction was made 
between farmers who stated to have created/developed the innovation by themselves 
(self-innovators) and those who declared to have learned of the innovation from exter-
nal sources. In this way, for the latter, the information channels can be explored in more 
detail by referring to a menu of possible sources. The external sources are split into three 
groups: institutional, market and acquaintances. The institutional group includes sources 
related to the sphere of agricultural research and extension, such as universities, research 
centres and other private and public entities (i.e. regional administrations, local authori-
ties, R&D from firms, training etc.); the market group refers to the sources of informa-
tion from producers, retailers and commercial agents; whereas the acquaintances group 
involves as a source of information the network of people surrounding each farmer, such 
as relatives, neighbours and others. This section seeks to highlight the role of informa-
tion and research, namely the elements representing the potential contribution to further 
understanding the innovation adoption process as well as the relative weight of agricul-
tural research to the farm-level effects of innovation adoption. The key element meant to 
establish a connection between external sources and effectiveness of the adopted innova-
tion is the investigation of the research-innovation link that is whether the farmers know 
about the research behind the development of the adopted innovation. 

The fifth section is dedicated to the declared effects of the introduced innovation 
in terms of changes in economic performance, combination of inputs and leisure time. 
Information on the effects on economic performance of the introduced innovation were 
collected by breaking down the profitability into four elements: Cost Reduction, Produc-
tion Increase, Value-Added Increase and Quality Increase. The importance of these vari-
ables within the context of the innovation adoption process is found in their potential to 
reveal the mechanism allowing the adopted innovation to contribute to the farms’ overall 
economic performance (profit). The sixth section investigates the future intentions of the 
farmers regarding the continuation of the agricultural activity and the adoption of further 
innovations in the next five years. Further inquiries are posed in order to record the opin-
ions of farmers about the relationship between innovation and agricultural policy, as well 
as the role of innovation for the improvement of competitiveness in agriculture. The last 
section concludes the questionnaire by inquiring into the future of the farm and of the 
farmers and eliciting opinions about innovation and the CAP. The data collected in this 
section are used for supporting the evaluation of, and better interpreting, some farmers’ 
choices, such as the motivations for not innovating.

Most of the data collected have been recorded as binary or categorical variables, 
whilst data related to farm size, labour, introduced innovations and others, have been 
recorded as continuous variables. Exceptions are represented by the information related 
to farms’ economic performance, which has been surveyed according to four variables, 



190 Michele Vollaro, Meri Raggi, Davide Viaggi

namely cost reduction, production increase and value-added increase, collected in per cent 
terms, and quality increase, recorded according to an ordinal categorical variable (not at 
all, low, high, very high). 

4. Case study area, data collection process and descriptive statistics

4.1 Case study area 

The agricultural sector in Emilia-Romagna is one of the most advanced and produc-
tive in Italy, due to the favourable geographical and climatic conditions (the southern part 
of the territory is mountainous, whilst the northern part belongs to the Po valley, which 
is a very fertile zone), and the presence of highly specialised enterprises. Emilia-Romagna 
is particularly active in the production of cereals (wheat and maize), fruit and livestock 
(mainly bovines, pigs and poultry) (Fanfani and Pieri, 2016). 

The Province of Bologna is located in the central part of the region, is agriculturally 
varied and composed of plains, hilly and mountain areas. According to the last agricultur-
al census carried out in 2010 by the Italian Institute of Statistics (ISTAT), the Province of 
Bologna accounts for about 10,800 agricultural units over an UAA2 of about 173,000 ha.

As shown in Table 1, the agricultural sector is mainly based on arable crops, involv-
ing about 7,000 farms and about 141,000 ha of UAA. Arable crop farming is mainly 
composed of farms growing cereals (about 4,000) and forage (about 2,000), whose UAA 
shares are 53% and 27%, respectively. The average size of farms producing cereals and for-
age crops is 12 and 10 ha, respectively, and more than half of them are located in plain 
areas. The second major type of farming in the province is livestock and related activities, 
involving about 800 cattle-holding farms with 33,000 heads as well as 150 swine-breeding 
farms and 75,000 heads. The largest livestock farms are based in plain areas. 

Regarding fruit cultivation, about 2,700 farms grow orchards over an UAA of about 
16,000 ha.

2 UAA stands for ‘Utilised Agricultural Area’.

Table 1. Agricultural census data per specialization (type of farming) and altitude level.

Specialization Plain Hill Mountain Total  

Cattle farms (Milk, Beef, ovine-caprine and mixed) 295 454 369 1118 10%
Cereal crops (wheat, maize, oats, barley) 3177 633 187 3997 37%
Other arable crops (open field, horticultural, mixed 
and grain pulses crops) 1284 849 608 2741 25%

Fruit (orchards, olives and grapes) 1529 1082 90 2701 25%
Non-classifiable 65 109 28 202 2%
Total 6350 3127 1282 10759  

59% 29% 12%    

Source: our elaboration on ISTAT data.



191Innovation adoption and farm profitability: what role for research and information sources?

4.2 Sample data and selected descriptive statistics

The sample, represented in Table 2, includes 178 farms located in the plains (59%), 
87 in hilly (29%) and 35 in mountain (12%) areas. According to the principal speciali-
sation, the sample is composed of 20 livestock farms, 116 cereal farms, 69 ‘other arable’ 
crop farms, 88 fruit farms (including olives, grapes and 11 nurseries), and 7 non-classifi-
able farms. Cereal crop results are the most frequent specialisation with about 39% of the 
total farms, followed by fruit farms (about 26%), arable crop farms (22%) and cattle farms 
(7%). Given that it is a direct result of the sampling procedure, the sample can be consid-
ered to be representative of the Province of Bologna.

Table 2. Sample units per specialisation (type of farming) and altitude level.

Specialisation Plain Hill Mountain Total  

Cattle farms (Milk, Beef, pork and mixed) 5 7 8 20 7%
Cereal crops (wheat, maize, oats, barley) 86 21 9 116 39%
Other arable crops (open field, horticultural, mixed 
and grain pulses crops) 33 21 15 69 23%

Fruit (orchards, olives and grapes) 49 36 3 88 29%
Non-classifiable 5 2   7 2%
Total 178 87 35 300  
  59% 29% 12%    

Source: our elaboration of primary data collected.

The sample accounts for about 8,000 ha of UAA, of which about 5,000 in ownership. The 
larger share of the land is that of cereal crop farms with about 36% of total land, followed by 
cattle farms (27%), (arable) crop farms (18%) and fruit farms (15%). The descriptive statistics 
of the collected data are presented in Table 1A (see Annex), while a wider presentation of the 
statistics of the sections from second to fifth is illustrated in the results section.

5. Results

5.1 Descriptive results

Altogether, 121 out of 300 farmers adopted at least one innovation in the last twenty 
years (about 40%) (the precise question was “in the last 20 years, what kind of product or 
process innovations have been introduced on your farm?”). This question was posed after 
asking the farmers about the existence of important innovations in agriculture (“Do you 
believe that in the last 20 years there have been very important innovations in your main 
field of specialisation (measured in terms of income)?”). Almost 47% (140 out of 300) of 
respondents replied positively. Table 3 provides an illustration of these results by crossing 
the answers to these two questions.



192 Michele Vollaro, Meri Raggi, Davide Viaggi

Table 3. Cross-tabulation of innovation introduction and consideration of important innovations in 
the last 20 years.

Introduction of at least one innovation in the last 20 years
Total

No Yes

Important innovations in the last 
20 years

No 125 35 160
Yes 54 86 140

Total 179 121 300

Source: own elaboration on collected data.

The consistent replies on the diagonal combinations (No-No and Yes-Yes) are some-
how intuitive, while a less straightforward reasoning may emerge from an analysis of the 
off-diagonal cross-answers. We discuss these four options, in turn, by also attaching some 
descriptive statistics of the farmers/farms belonging to each combination.

The 125 No-No answers (roughly 42% of the sample) are composed of 50% cereal, 
26% other arable crop and 15% orchard growers. With respect to the total of each spe-
cialisation, cereal growers represent 54% (63 out of 116), other arable crop, 49% (33 out of 
67) and the orchard growers, 25% (19 out of 77). These 125 respondents are mainly small 
farms with low agricultural income. In fact, on average, 83% of them operate on less than 
20 hectares and 76% of them have an income from agricultural activities that accounts for 
less than 30% of family income. Such conditions are consistent with the declared reasons 
for ‘no adoption’, mainly related to high costs.

The 54 Yes-No answers indicate no adoption in spite of the existence of important 
innovations in the sector of specialisation. These 54 farms are composed of 41% cereal, 
24% other arable crops and 30% orchard growers. With respect to the total of each spe-
cialisation, cereal growers represent 19% (22 out of 116), other arable crop, 19% (13 out 
of 67) and the orchard growers, 21% (16 out of 77). This group also is composed of small 
farms with low agricultural income, but the frequency of these types of farms is slightly 
lower than in the previous group. In fact, on average, 72% of them operate on less than 
20 hectares and 71% receive an income from agricultural activities that is less than 30% of 
family income. In this case also, such conditions seem to be consistent with the declared 
reasons for ‘no adoption’, mainly related to high costs, the expectation of soon retiring 
from farming (cereal) and maintaining production traditions (orchard). 

The 35 No-Yes replies indicate adoption despite the declaration that there have been 
no important innovations in the sector of specialisation. These 35 farms are composed of 
17% livestock farms, 43% cereal producers, 9% of other arable crops and 26% of orchard 
growers. With respect to the total of each specialisation, livestock farms represent 30% (6 
out of 22), cereal growers, 13% (15 out of 116), other arable crops, 4% (3 out of 67) and 
the orchard growers, 12% (9 out of 77). This group of innovators is characterised by the 
fact that they operate on larger farms with higher agricultural incomes. In fact, on aver-
age, only 47% of them operate on less than 20 hectares and 49% receive an income from 
agricultural activities that is less than 30% of family income. These figures clearly differ 
from those of the previous two groups and the declared motivations for having introduced 



193Innovation adoption and farm profitability: what role for research and information sources?

at least one innovation (with positive effects on profitability) mostly refer to reducing 
costs and increasing production. Both groups, and in particular the first one, are consist-
ent with innovations that are more linked to late adoption of existing solutions motivated 
by economies of scale, rather than by strong innovation behaviour.

In the last group, the 86 Yes-Yes replies consist of 12% of livestock farms, 19% of 
cereal producers, 21% of other arable crops and 38% of orchard growers. With respect 
to the total of each specialisation, livestock farms represent 45% (10 out of 22), cereal 
growers 14% (16 out of 116), other arable crops 28% (18 out of 67) and the orchard 
growers represent 43% (33 out of 77). This other group of innovators operates on 
farms with sizes similar to the ones of the previous group, but with higher agricultural 
income. In fact, on average, 43% of them operate on less than 20 hectares and only 27% 
of them obtain an income from agricultural activities that is less than 30% of family 
income. Similarly to the previous group, this last group also indicates as the main moti-
vations for adopting at least one innovation (with positive effects on profitability) cost 
reduction and production increases, with the addition of other motivations pertaining 
to the improvement of labour conditions, such as reducing fatigue and improving the 
safety of workers. This profile, which is particularly consistent with the orchard speciali-
sation, denotes farmers who are focusing on agricultural production on well-structured 
farms and who are open to an understanding of the outside markets’ trends as well as 
who are highly focused on innovation.

On the other hand, about three-fifths of the interviewees (179 farmers) decided not 
to innovate due to the economic and managerial hurdles that reduce the capacity of farm-
ers to obtain new technology and adopt innovations3. We asked these farmers to motivate 
their decision not to adopt innovation by choosing among two categories of responses: 
obstacles and intentional choice. Among the obstacles, we proposed high costs, bureau-
cracy and risks, while for intentional choice we asked about ethical reasons, the intention 
to quit the business, negative past experiences and the desire to maintain traditional pro-
duction processes. Eighty-four (84) out of 179 replies deemed the excessive costs of adopt-
ing innovations to be the main hurdle, while 16 and 18 answers indicated their intention 
to quit the business soon and to keep maintain production traditions, among the inten-
tional choice group, respectively. Therefore, the sample revealed that the main reason for 
not having adopted innovations in the last 20 years was the excessive cost, highlighting 
economic barriers and the lack of managerial skills for gaining access to new technolo-
gy. However, since for 33 out of 51 (65%) other reasons were expressed by cereal farms, 
we deduce that for about two-thirds of respondents such choice is due to a disinterest 
in innovation given that they possess less than 20 hectares, no weeding and harvesting 
machines, and therefore opt for the services of other companies. This is an important 
point considering the recent structural trends as it points at a dichotomy between larger 
professional farms, for which innovation remains important, and small farms that keep 
land tenure but farm via contracts, for which innovation is rather carried out or adopt-
ed by contractors themselves, i.e. outside the farm. For the remaining third, we equally 
deduce that they have not been interested in adopting innovations, but unlike the pre-

3 Detailed descriptions of such data have been omitted in order to save text. These are, however, available from 
the authors upon request.



194 Michele Vollaro, Meri Raggi, Davide Viaggi

vious farmers, because the technology they possess is considered to still be effective and 
hence does not need to be replaced or upgraded.

The number of innovations introduced in the last 20 years is more than 200 for 121 
innovators (an average rate of about 2 innovations per farmer). 

The distribution of adoptions, shown in Table 4, reveals that mechanical innovations 
are the most frequently adopted ones (32%), followed by energy-water saving (21%), 
diversification (15%) and biological, agricultural and informatics (about 8% each). The 
distribution of type of innovations changes if considering the unique (most important) 
innovation that, according to farmers, yielded the highest impact on profitability. In fact, 
the shares of mechanical (42%) and energy-water saving (25%) innovations increases, 
while the others decreased slightly. As for motivations, the adoption of these types of 
innovations is mainly motivated by the need to reduce costs, to increase production and 
to face new climatic challenges affecting the availability of natural resources, such as water.

As concerns the timing of introduction, about 65% of the mechanical innovations 
were introduced during the 2010-2015 period, while about 66% of the energy-water sav-
ing technologies were adopted in the 2005-2015 period. The adoption timing of the other 
types of innovation is smoothly spread across the considered time span (1995-2015).

The main reasons motivating the adoption of the (one) most important innovation 
are concentrated in cost reduction (66 or 35%) and production increase (56 or 30%) (122 

Table 4. Number of innovations introduced in the last 20 years and selection of the most important in 
terms of profitability.

Type of adopted innovations All adoptions Share of adoptions 
on total

Unique adoption 
considered most-

important in 
terms of impact 
on profitability

Share of important 
innovations on 

total

Biological-Genetic 18 8.5% 8 7.5%
Diversification or 
Manufacturing 32 15.0% 15 14.0%

Agricultural-Zootechnic 18 8.5% 7 6.5%
Mechanical-Automation 68 31.9% 45 42.1%
Informatics 17 8.0% 2 1.9%
Energy-Water saving 
(irrigation plants, solar panels, 
biogas) 

44 20.7% 27 25.2%

Marketing strategies 
(quality systems, production 
protocols) 

5 2.3% 2 1.9%

Operational (cooperatives, 
associations, logistics) 2 0.9% 0

Other 9 4.2% 1 0.9%
Total adoptions 213 100% 107 100%
Does not know 14

Source: own elaboration on collected data.



195Innovation adoption and farm profitability: what role for research and information sources?

replies out of 187)4. However, out of these 122 replies, 31 prove to be jointly chosen by the 
same farmer, indicating an important synergy between the two aspects in contributing to 
the increase of profitability5. Other motivations, collected in open format, result in gen-
eral profitability improvement, without any reference to specific motivation, and reduc-
tion of worker fatigue. The main motivations for cost reduction and production increases 
are more frequent for cereals (25%), fruit (19%) and grape farms (16%). In particular, 
by looking at the (one) most important innovations, mechanical-automation and ener-
gy-water saving proves to be the most frequent with 32 and 20 replies out of 66 for cost 
reduction and 19 and 10 out of 56 for increasing production, respectively. 

Beyond the motivations underlying the choice of the selected innovations, the sur-
vey investigated the selection and the adoption processes operated by the farmers. Indeed, 
farmers were asked whether they designed and/or developed the (adopted) innovation by 
themselves or obtained the information regarding the introduced innovations from exter-
nal sources (and from whom the farmer was informed about the existence of such innova-
tion).

In this respect, farmers who declare to have designed and/or developed the innova-
tion by themselves are denominated “self-developers” and are considered to be the(ir) 
internal source of information as opposed to the other innovators who declared to have 
learned about the innovation from an external source of information. 

The data about the sources of information, shown in Table 5, indicate self-developed 
innovation in the first column and the list of proposed external sources. Self-development 
of innovation has been declared by 31% of innovators, with prevalence for cereal, fruit 
and nursery farms. It follows that the remaining 69% learned about the innovation from 
external sources and, in particular, mostly from sources other than public institutions 
and unions/farmer associations. Indeed, 37% of the innovators declared to have acquired 
information about the innovation they decided to introduce from consultants, courses, 
local and visits to farms abroad. The second largest share is the 17% represented by the 
sources of information from people belonging to the sphere of personal relationships of 
the farmers such as friends, relatives and neighbours. 

Unions and sectorial associations cover 10% of the external sources of information 
and the relative frequency appears to be uniformly distributed across specialisations. Only 
a residual share of about 2% represents the public institutions devoted to research and 
development in agriculture as the external sources of information. Such a result highlights 
the importance of intermediation between research and farmers. 

As a follow up question, farmers were asked to declare their knowledge of the maker/
producer of the innovation. By excluding self-developers, this inquiry reveals that most 
innovators (about two-thirds), who learnt about the existence of the introduced innova-
tion from external sources, were also aware of who developed the innovation. This might 
indicate that farmers engage in a careful decision-making process before adopting the 
innovation or at least show a good level of awareness about its background. Qualitative 

4 The number of replies is greater than the number of adopters as the inquiry was devised as a multiple-choice 
question.
5 The link was not explicitly asked, but, in the explicit list, we included the reduction of risks and the diversifica-
tion of the activity in order to evaluate the motivations directly related to profitability. Very few replies were col-
lected.



196 Michele Vollaro, Meri Raggi, Davide Viaggi

Ta
bl

e 
5.

 S
ou

rc
es

 o
f i

nf
or

m
at

io
n 

fo
r i

nn
ov

at
io

n 
ad

op
te

d 
pe

r s
pe

ci
al

iz
at

io
n.

Sp
ec

ia
liz

at
io

n
Se

lf 
de

ve
lo

pe
d

Ex
te

rn
al

 so
ur

ce
 o

f i
nf

or
m

at
io

n

N
o 

re
pl

y
To

ta
l

In
st

itu
tio

ns
(U

ni
ve

rs
ity

, R
eg

io
n,

 
Pr

ov
in

ce
, M

in
ist

ry
)

U
ni

on
s, 

as
so

ci
at

io
ns

A
cq

ua
in

ta
nc

es
, 

fr
ie

nd
s, 

re
la

tiv
es

, 
ne

ig
hb

ou
rs

O
th

er
 so

ur
ce

s 
(c

on
su

lta
nt

s, 
re

fr
es

he
r c

ou
rs

es
/

tr
ai

ni
ng

s, 
vi

sit
in

g…
)

M
ilk

-b
ee

f c
at

tle
Be

ef
 c

at
tle

1
1

2
4

M
ilk

 c
at

tle
3

2
2

2
9

M
ix

ed
 c

at
tle

, m
ai

nl
y 

pa
st

er
n

1
1

O
vi

ne
-c

ap
rin

e 
an

d 
pa

st
er

n 
ca

ttl
e

1
1

C
er

ea
l c

ro
ps

 (w
he

at
, m

ai
ze

, o
at

s, 
ba

rle
y)

12
3

6
10

31
O

pe
n 

fie
ld

 c
ro

ps
2

1
1

4
7

M
ix

ed
 c

ro
ps

H
or

tic
ul

tu
ra

l c
ro

ps
2

2
2

3
1

10
H

ig
h 

pr
ot

ei
n 

cr
op

s (
gr

ai
n 

pu
lse

s)
2

1
3

C
om

bi
na

tio
n 

of
 c

ro
ps

 a
nd

 c
at

tle
1

1
Fr

ui
t 

9
1

2
4

9
25

O
liv

es
1

1
G

ra
pe

s
3

1
2

10
16

N
ur

se
ry

5
1

2
8

N
on

-c
la

ss
ifi

ab
le

1
1

1
3

To
ta

l
38

2
12

20
45

4
12

1
31
.4
%

1.
7%

9.
9%

16
.5

%
37
.2
%

3.
3%

So
ur

ce
: o

w
n 

el
ab

or
at

io
n 

on
 c

ol
le

ct
ed

 d
at

a.



197Innovation adoption and farm profitability: what role for research and information sources?

additions during the interview revealed indeed that farmers rely upon trusted external 
sources of information and acquaintance with the producers6,7. Overall, the sample reveals 
that the majority of farmers either strictly rely on their own ability to develop an innova-
tion or, on one’s own initiative, search for information and cues from others’ experience in 
order to make the best innovation choice and to meet their profit expectations. 

In order to explore the connection between innovation adoption at farm level and 
research, farmers were asked to state whether they knew that the innovation they adopted 
originated from a specific agricultural research. This question was addressed only to those 
farmers that previously declared to have learned of the innovation through an external 
source of information. We excluded self-innovators from this question because we sup-
pose that they engage in a process for introducing innovation that is mainly based on the 
self-development of their own ideas, which is completely different from the process fol-
lowed by the other interviewed innovators. Hence, this question was asked to 83 innova-
tors. Fifty-three respondents (about 64%) stated that they knew that the innovation was 
derived from specific research in agriculture. In particular, 29 out of these 53 (about 55%) 
concern mechanical innovations, mainly related to cereal, grape and fruit farms.

The stated effects on economic performance are reported in Figure 1. Cost Reduction, 
Production Increase, and Value-Added Increase are measured in per cent increase, while 
Quality Increase is measured through four categorical levels (not at all, low, high and very 
high) of increase due to the introduction of the innovation. 

The number of observations of these variables does not correspond to the numerous-
ness of the innovators’ sub-sample (121), because not all respondents provided a reply to 
each of the four questions. Zero answers correspond to the actual observation of the per-
formance by the farmer, while a missing reply might be justified by the lack of expecta-
tion, detection or perception of any impact on that specific component of profitability (in 
fact many farmers stated to not know the specific performance). Since the answers were 
not mutually exclusive, respondents had the choice to indicate more than one positive 
effect and potentially all of the four asked.

Cost Reduction (A), Production Increase (B), Value-Added Increase (C) show a note-
worthy frequency of zeros; this was expected since it is unlikely that one innovation 
might yield positive profitability outcomes on all of the four considered components at 
the same time. The effect on Cost presents a concentration of positive outcomes within 
the range of 10-60% cost reduction (with the highest share on the lower boundary of 
the interval and no case recorded between 40% and 50%), while Production and Value-
added are more frequently within the 10-40% interval of increase. Production Increase 
also shows a fairly high frequency around the 50-60% range. As far as Quality increase 
is concerned, it is observed that about 60% of the replies indicate an improvement in 
profitability due to high and very high quality increases, while only about 25% show no 
quality increases at all. 

6 For some types of innovation, such as mechanical ones, farmers have a better knowledge of the major brands/
producers because of the presence, in the Emilia-Romagna region, of a large number of mechanical manufactur-
ers that have been operating there since the beginning of the last century. Farmers in the Province of Bologna 
possess a deep knowledge of the evolution of mechanical technologies and mechanical manufacturing, which 
provides them with a sufficient ability to develop their own mechanical innovations.
7 Detailed results are available from the authors upon request.



198 Michele Vollaro, Meri Raggi, Davide Viaggi

5.2 Econometric analysis

According with the methodology illustrated in section 3, the results obtained from 
the econometric analyses are reported in two groups: the first pertains to the adoption 
of innovation (adoption models) and the second concerns the linkage between adopted 
innovation and performance (performance models). 

The results of the Poisson and Probit adoption models, shown in Table 6, indicate 
which factors are most important in determining respectively the number of innovations 
and the choice of (propensity to) introducing an innovation.

The ability of both models to analyse the survey data is quite good, as indicated by 
the Wald χ2 statistics. The results from both models indicate that the propensity to inno-
vate, in particular to adopt more than one innovation, is highly determined by the eco-

Figure 1. Frequency distribution of Cost Reduction (A), Production Increase (B), Value-Added Increase 
(C) and Quality Increase (D).

A (63) B (71) 

  

C (75) D (121) 

  

 1 
Source: own elaboration on collected data; number of observations in parentheses.



199Innovation adoption and farm profitability: what role for research and information sources?

nomic size and other structural characteristics of the farm, as well as by some individual 
and behavioural characteristics of the respondents. The positive role of the share of rented, 
over total, land may be connected to both the structural characteristics of the farm, like-
ly qualified by a rent-based expansion, and to the overall size in terms of land area. The 
number of tractors is positively and significantly correlated to the number of innovations 
(but not to the decision to innovate) and shows that multiple innovations are more likely 
on large and capital-intensive farms. The positive and significant coefficient of the share of 
agricultural income shows a higher propensity to innovate on more professionally farms 
focused on agricultural activity. On the contrary, a higher number of family labourers and 
the juridical status of individual farms indicate that small farms are less inclined to adopt 
innovation (these are also correlated to the specialisation given the remarkable share of 
small cereal farms). As concerns individual and behavioural features, instead, we can 
observe that more educated farmers and those declaring that, in the last 20 years, impor-
tant innovations in terms of profitability have been released show a higher propensity to 
innovate and, in particular, to adopt more than one innovation. 

In order to further support these first results, and to better explain the process, a two-
step model has been applied by employing a double-hurdle regression8. The results are 
shown in Table 7.

8 Thanks to an anonymous referee for the suggestion of including a two-step model.

Table 6. Poisson and Probit adoption models.

Characteristics

Number of introduced 
innovations

(Poisson)

Introduction of 
innovation (0-1)

(Probit)

Coefficient Marginal 
effect Coefficient Marginal 

effect

Innovation Important innovations (last 20 yrs.) 0.91*** 0.66*** 0.78*** 0.21***

Farm Share of rented over total land 0.73*** 0.53*** 1.00*** 0.26***

Number of tractors 0.05*** 0.03*** 0.03 0.01
Livestock specialisation 0.22 0.16 0.42 0.11
Cereal specialisation -0.54*** -0.39*** -0.59*** -0.16***

Socio-economic Education > than mid-school 0.64*** 0.47*** 0.57*** 0.15***

Family income from Agric<30% -0.58*** -0.42*** -0.53*** -0.14***

Number of family labour -0.16** -0.11** -0.11 -0.03
Individual farm -0.38* -0.28* -0.59*** -0.16***

Constant -0.70**

Observations 244 244
Wald χ2 146*** 80***

AIC 478.6 248.4
BIC 513.6 283.4

Note: robust standard errors; * p < 0.10, ** p < 0.05, *** p < 0.01.



200 Michele Vollaro, Meri Raggi, Davide Viaggi

Table 7. Double-Hurdle model.

Characteristics Number of introduced 
innovations

Outcome (quantity) equation
Farm Number of tractors 0.09**

Livestock specialisation 1.15**

Cereal specialisation -0.64*

Fruit specialization, including grape and olives 0.29
Socio-economic Specialised Ag education -0.47*

Family income from Ag <30% -0.63**

Family workers per ha 1.00*

Constant 1.21**

Choice (participation) equation
Innovation Important innovations in last 20 yrs 1.01***

Farm Location: plain=1; hill=2; mountain=3 -0.22**

Total Land 0.01***

Socio-economic Education superior than middle school 0.75***

Family workers per ha -0.65**

σQ 1.86***

σQσP -1.69***

Observations 245

Note: robust standard errors; * p < 0.10, ** p < 0.05, *** p < 0.01; σQ is the estimated value of the stand-
ard deviation of the error term of the quantity equation; σQσP is the estimated value of the covariance 
between the error terms of the quantity equation and the participation equation.

The results obtained through the double-hurdle model confirm those from the Poisson 
and the Probit models. In addition, they indicate that the choice of innovating depends 
highly upon location, especially in the plains and hills. Larger farms and higher education 
contribute to improve the probability of adoption. The consideration of important innova-
tions in the last 20 years notably affects adoption, but it does not contribute to explain 
the number of adoptions. Moreover, what seems to determine increases in the number 
(quantity) of adoptions are factors related to the type of farming (and relative physical and 
economic size of the farm). In fact, larger farms with higher agricultural income, such as 
livestock farms, or farms with higher family labour and higher mechanisation (number of 
tractors) are more prone to adopt more than one innovation.

The core part of the analysis concerns the explanation of the economic performance 
of the adopted innovation, specifically in relation to its origin from research and in con-
nection to the source of information. The results from the probit performance model con-
cerning each of the four components of the farm’s profitability are shown in Table 8. 

Given the application of the performance models to each measure of performance, the 
number of observations for each group of regression is reduced with respect to the entire 
sample.



201Innovation adoption and farm profitability: what role for research and information sources?

Table 8. Probit performance models.

Characteristics

Economic performance

Cost reduction 
[yes=1; no=0]

Production 
increment 

[yes=1; no=0]

Value added 
increment

[yes=1; no=0]

Quality 
increment 
[very high, 

high=1; 
otherwise=0]

Innovation Research-innovation link 0.21 0.77* 0.76* 0.83**

Source of innovation 
[external=1; self =0] -0.93 -1.22* -1.59** -0.87**

Age of innovation -0.01 0.04 0.06** 0.05*

Important innovations 
(Last 20 years) -0.73 -1.13** 0.05 -0.07

Farm Cereal specialisation 0.65 0.53 -0.67 -0.05
Share of rented land over total 
land -0.42 -0.27 -0.42 0.18

Socio-
economic

Individual farm [yes=1; no=0] -0.20 0.11 0.67* 0.10
Family income from Ag <30% -0.91* -1.72*** -1.37*** -0.32
Education > than mid-school 0.34 -0.11 0.52 -0.28
Constant 1.98** 1.88** 0.25 0.41

Observations 50 56 62 88
Pseudo R2 0.176 0.245 0.317 0.115
Wald χ2 12.1 14.9* 30.4*** 11.7
AIC 71.7 76.6 78.7 123.9
BIC 90.8 96.9 100.0 148.7

Note: robust standard errors; * p < 0.10, ** p < 0.05, *** p < 0.01.

The probit performance models applied on cost reduction and quality improvement 
proved to have a scarce capacity to explain the likelihood of obtaining positive perfor-
mances. In the first model (cost reduction) only one regressor out of nine is significant and 
the sample is relatively small, while in the last model only the group of information vari-
ables contributes to explaining the variability in quality improvement. 

On the contrary, the probit performance model proved to perform better when applied 
on production and value-added increment. In fact, for the latter models, the results show 
significant contributions in both groups of variables. From all significant results, a com-
mon pattern can be identified in the positive contribution of innovation originating from 
research, but also in the negative effect of external information on the likelihood of obtain-
ing a positive economic performance.

These results indicate that farmers who knew the innovation from external sources 
have lower chances to obtain positive economic performance, especially in terms of value-
added and production, with respect to self-innovators. On the other hand, the positive con-
tribution of research on economic performance is more pronounced in terms of quality. 



202 Michele Vollaro, Meri Raggi, Davide Viaggi

However, although the probit performance analysis provides interesting results, its 
specification might be affected, beyond the reduced number of observations, by selec-
tion bias in that only farmers who expect higher economic performance, on the basis of 
the information they possess, might decide to effectively adopt the innovation. In order 
to evaluate such a hypothesis, a Heckit model, specifically a probit model with sample 
selection, is run by formally dividing the variables into two groups, namely the selection 
(adoption) and outcome (performance) variables. The Heckit models indicate the presence 
of a self-selection process of innovation introduction related only to positive expected 
gains in value-added, as indicated by the significance of ρ, while the other model specifica-
tions indicate that both processes are essentially independent9 (Table 9). 

The results indicate that the Heckit models appear to be more appropriate in explain-
ing the effects of the information variables on the economic performance. Indeed, these 
models, on one hand, confirm the results related to research and source of information 
from the previous probit performance models, and, on the other hand, report the same 
results as the introduction models, with the exception of the variable number of tractors.

6. Discussion

In this paper we investigate the determinants of innovation adoption and the relation-
ship between origin of innovation and economic performance at farm level.

In the sample considered there is a noteworthy share of farmers who are actively 
innovating, which is partly explained by the long-time horizon taken into account. Most 
frequent innovations are in the field of mechanical innovations and innovation aimed 
at water-energy saving. This is consistent with the fact that mechanisation is a wide-
spread need across farm specialisations, on the one hand, and with the current need to 
save resources in a context characterised by climate change; the latter issue is potentially 
emphasised by the location of the study area in a Mediterranean region. Multiple innova-
tions are frequent among innovators, which may be explained by both the existence of 
connections among innovations (innovation packages) and the tendency of most active 
farm(er)s to innovate continuously (Läpple et al., 2015).

The results from the adoption models, mainly testing the adoption determinants, are 
largely consistent with the findings in literature in terms of structural characteristics of the 
farms, such as farm size, mechanization, labour and production type, and subjective char-
acteristics of the farmers, such as farmer education, experience and off-farm income10. 
The main novelty arises from the consideration of the judgement of farmers regarding the 
existence of important innovations in their field of specialisation, which helps to distin-
guish between cases in which the innovation choice by the farm results from the need of 
keeping up with a general technology shifts (i.e. replacing obsolescence), aligned to the 
technological treadmill, from cases in which innovation is more a choice tuned to the spe-
cific production and marketing needs of the farm. It also helps to understand the differ-

9 Indeed, results were verified by running a probit regression on the performance variables by solely employing 
the information variables. The results confirm the ones obtained in the output equation of the Heckit model.
10 The consistency of our results has been compared to the following literature: Feder and Slade, 1984; Lin, 2001; 
Daberkow et al., 2003; Diederen et al., 2003; Dimara et al., 2003; Kounduri et al., 2006; Cavallo et al., 2014, 
Läpple et al., 2015; Ramos-Sandoval et al., 2018; Sauer et al., 2019.



203Innovation adoption and farm profitability: what role for research and information sources?

ent profiles of the non-innovators, namely those for whom no-innovation is linked to the 
absence of innovation in the sector in contrast to those foregoing innovation for personal 
or farm reasons, in spite of the progresses of innovations in the sector.

The second group of models, namely the performance models, represent, in our 
knowledge, the first attempt to evaluate the existence of a relationship between research 
and farm performance, also taking into account farmer intermediation. The first results 
support the hypothesis of a differential impact of innovations originating from research, 
which increase profitability by positively affecting value-added and quality improve-

Table 9. Probit performance model with sample selection.

Characteristics

Economic performance

Cost reduction 
[yes=1; no=0]

Production 
increment 

[yes=1; no=0]

Value added 
increment 

[yes=1; no=0]

Quality 
increment 
[very high, 

high=1; 
otherwise=0]

Outcome equation (O)
Innovation Research-innovation link 0.31 0.53 0.58* 0.79**

Source of innovation 
[external=1; self =0] -1.20* -1.16** -1.16*** -0.91**

Age of innovation -0.01 0.03 0.04 0.05*

Constant 1.36** 0.98** 0.98** 0.30

Selection equation (S)

Innovation Important innovations (last 20 
years) 0.41* 0.48* 0.48** 0.84***

Farm Breeder specialisation 0.52 -0.11 -0.20 0.56
Cereal specialisation -0.47** -0.32 -0.64*** -0.52**

Share of rented over total land 1.15*** 0.76** 0.99*** 1.03***

Number of tractors 0.04 0.05 0.05 0.02

Socio-
economic Education > than mid-school 0.44* 0.49** 0.53*** 0.52**

Family income from Ag <30% -0.22 -0.62*** -0.44** -0.55***

Family labour -0.26** -0.29*** -0.22** -0.09
Individual farm 
[yes=1; no=0] -1.13*** -0.75*** -0.77*** -0.71***

arctan(ρ)† 0.05 -0.27 -1.13* -0.14
Observations 241 243 240 232
Uncensored Obs 50 56 62 88
AIC 272.5 301.6 308.7 344.3
BIC 321.3 350.5 357.4 392.5
Wald χ2 (O) 3.76 4.55 8.42** 10.2**

Note: robust standard errors; t statistics in parentheses; * p < 0.10, ** p < 0.05, *** p < 0.01; † arctan(ρ) 
indicates the correlation coefficient between output and selection equations.



204 Michele Vollaro, Meri Raggi, Davide Viaggi

ments. On the contrary innovations originating from research do not appear connected to 
improvements in productivity or cost reduction.

Although this paper contributes to evidence on the role of research and information 
sources in improving farms’ economic performance, it is also affected by some limitations 
that may affect the robustness and the generalisation potential of the results. First, the 
sample is rather small, in particular for the adopters’ subsample, in particular consider-
ing the heterogeneity brought about by the large coverage of different farm specialisations. 
This may have contributed to the low significance of some of the models and some diffi-
culty in estimation. This has also made potential additional explanatory variables difficult 
to use.

Second, the case study relies on a specific province in Italy, which, while benefiting 
from an internal heterogeneity (in terms of farm specialisation and altitude), still repre-
sents a specific context in terms of general ecological and legal conditions (including spe-
cific priorities e.g. for investment).

A third limitation concerns the way the data were collected. Due to a lack of better 
information availability (e.g. from accounting data) and resource limitations, most of the 
variables are based on statements made by farmers. This is a sensible topic, in particu-
lar with respect to the estimation of the impact of innovation on profitability parameters, 
which also implies a request for a difficult judgement on the part of the farmers, and of 
the origin of innovation, especially with respect to research, that incorporates a mix of 
actual information about the origin and level of documentation by the farmers. The origin 
of innovations and knowledge about it, in turn, relate to each other and are almost impos-
sible to distinguish in the way in which the survey was run. Based on other questions and 
statements by farmers on their own level of information, we can interpret this information 
mostly as revealing the true origin of innovation, however there is certainly some level of 
(unmeasurable) approximation. 

Fourth, and connected to the above, using stated information coupled with resource 
constraints implied the need to collect this information in a simplified way (e.g. using 
qualitative or dichotomous variables) and, in some cases, to use classes in the data treat-
ment in order to account for “perceptive discontinuities” (such as round numbers in per 
cent statements). This, however, implies some further difficulty in the estimation and 
interpretation of the models.

These limitations, associated with the promising results achieved, highlight relevance 
and provide more precise hypotheses for further investigation on this issue. This would 
require, however, a larger sample, wider territorial coverage and would benefit from link-
ages to structural and performance data not available for this study.

An important message arising from the paper, in spite of the limitations, is that the 
role of farmers is crucial for innovation development and that farmers who are willing 
to innovate are engaged in a continuous learning process which includes, beyond the 
practical knowledge of the available innovations, the knowledge and awareness of the 
process leading from research to the realisation of the innovation as well. This evidence 
supports the paradigmatic change of the innovation process from AKS towards the AKIS 
and multi-actor concepts (SCAR, 2012), by providing additional insight into the proac-
tive role of farmers in the management of external information coming from different 
sources, including research, and of own-knowledge within the innovation adoption pro-



205Innovation adoption and farm profitability: what role for research and information sources?

cess (Klerkx et al., 2009; Läpple et al., 2015). Such proactivity might represent a relative 
competitive advantage for the improvement of farm performance and a key feature of 
entrepreneurship. However, its ‘anatomy’ would need to be better analysed in future stud-
ies, with the collection of more specific information about on-farm processes leading to 
innovation adoption or implementation on the farm. 

7. Conclusions

The results of this paper show the importance of innovation for a large share of farms, 
considering a substantial time frame of 20 years. Most frequent innovations are in the 
field of mechanical innovations and innovation aimed at water-energy saving. Multiple 
innovations are frequent among innovators.

Classical factors, such as proxies related to farm size, remain the most suited varia-
bles to explain the adoption of innovations, while motivations for innovation adoption are 
largely related to the combination of cost reduction and production increases.

The process of innovation development and adoption follows two main pathways: 
self-development by farmers and development by mostly private companies. Agricultur-
al research is generally known to be in the background, but rarely seems to lead direct-
ly to technology development and even less to adoption. This may also be connected to 
the prevailing technologies that are considered to be relevant in the area (mechanisation 
and water/energy saving), which require important steps in terms of ‘engineerisation’ of 
knowledge and fine tuning in local conditions (including machinery set-up and feedback 
from users). In either case, the mediation between research and farmers has an important 
industry component or, in any case, involves different layers of actors.

The (knowledge of) existence of research activities in developing the innovation 
seems to be associated to better performance only for the specific but important cases 
of improving the value-added and of achieving very high-quality production. This sug-
gests that scientific research can have a specific role in terms of different performance-
improving strategies, and, in particular, that it can contribute comparatively more to qual-
ity, while self-development or industry-led technology adaptation can have a better role in 
cost reduction.

These results also yield relevant insights in terms of research policy. In particular, 
when promoting multi-actor approaches, innovation policies should better consider dif-
ferent regional/sector objectives in terms of quality, productivity or cost reduction, and 
related to this, more explicitly evaluate the potentially different roles of private and pub-
lic research and innovation players. In addition, while it can be expected that economic 
incentives linked to factor and product prices mainly affect cost reduction through self-
innovation, a stronger role has anyway to be attributed to direct research and innovation 
incentives if quality objectives are to be pursued.

In spite of its limitations, the study hints at the need to further explore the co-exist-
ence and interplay among different innovations, different innovation pathways and dif-
ferent innovation impacts. Moreover, the interaction between awareness of technology 
development pathways and actual technology performance at farm level is an issue that 
was only partially untangled in this paper and one that is undoubtedly worthy of further 
investigation. 



206 Michele Vollaro, Meri Raggi, Davide Viaggi

8. Acknowledgement

This study was conducted in the framework of the “IMPRESA” project, which 
received funding from the European Community’s Seventh Framework Programme under 
the GA 609448. We would like to thank Pedro Andres Garzon Delvaux and Pavel Cia-
ian for the review of a previous version of the paper.  The content of this study does not 
reflect the official opinion of the European Union. Responsibility for the information and 
views expressed in the present paper lies entirely with the authors.

9. References

Besley, T. and Case, A. (1993). Modeling technology adoption in developing countries. 
The American Economic Review 83(2): 396-402.

Cavallo, E., Ferrari, E., Bollani, L. and Coccia, M. (2014). Attitudes and behaviour of 
adopters of technological innovations in agricultural tractors: A case study in Italian 
agricultural system. Agricultural Systems 130: 44-54.

Cochrane, W.W. (1958). Farm prices: myth and reality. U of Minnesota Press.
Daberkow, S.G. and McBride, W.D. (2003). Farm and operator characteristics affecting the 

awareness and adoption of precision agriculture technologies in the US. Precision 
Agriculture 4(2): 163-177.

Davis, F.D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of 
information technology. MIS Quarterly 13(3): 319-340.

Diederen, P., Van Meijl, H., Wolters, A. and Bijak, K. (2002). Innovation adoption in agri-
culture: innovators, early adopters and laggards. Cahiers d’Economie et de Sociologie 
Rurales 67: 29-50.

Diederen, P., van Meijl, H. and Wolters, A. (2003). Modernisation in agriculture: what 
makes a farmer adopt an innovation? International Journal of Agricultural Resources, 
Governance and Ecology 2(3-4): 328-342.

Dimara, E. and Skuras, D. (2003). Adoption of agricultural innovations as a two‐stage par-
tial observability process. Agricultural Economics 28(3): 187-196.

Esposti, R. (2012). Knowledge, technology and innovations for a bio-based economy: lessons 
from the past, challenges for the future. Bio-based and Applied Economics 1(3): 235-268.

Fanfani, R. and Pieri, R. (Eds.) (2016). Il sistema agro-alimentare dell’Emilia-Romagna. 
Osservatorio Agro-Alimentare, Unioncamere e Regione Emilia-Romagna Assessor-
ato Agricoltura, caccia e pesca.

Feder, G. and Slade, R. (1984). The acquisition of information and the adoption of new 
technology. American Journal of Agricultural Economics 66(3): 312-320.

Feder, G. and Umali, D.L. (1993). The adoption of agricultural innovations: a review. Tech-
nological Forecasting and Social Change 43(3): 215-239.

Flett, R., Alpass, F., Humphries, S., Massey, C., Morriss, S. and Long, N. (2004). The tech-
nology acceptance model and use of technology in New Zealand dairy farming. 
Agricultural Systems 80(2): 199-211.

Folorunso, O. and Ogunseye, S.O. (2008). Applying an enhanced technology acceptance 
model to knowledge management in agricultural extension services. Data Science 
Journal 7: 31-45.



207Innovation adoption and farm profitability: what role for research and information sources?

Foster, A.D., and Rosenzweig, M.R. (1995). Learning by doing and learning from others: 
Human capital and technical change in agriculture. Journal of Political Economy 
103(6): 1176-1209.

Ghadim, A.K.A. and Pannell, D.J. (1999). A conceptual framework of adoption of an agri-
cultural innovation. Agricultural Economics 21(2): 145-154

Hayami, Y. and Ruttan, V. (1985). Agricultural Development: An International Perspective. 
Baltimore: Johns Hopkins University Press.

Hall, A. (2012). Partnerships in agricultural innovation: Who puts them together and are 
they enough? In OECD (Eds.), Improving Agricultural Knowledge and Innovation 
Systems: OECD Conference Proceedings, OECD Publishing, 219-236.

Hockmann, H., Garzon Delvaux, P.A., Voigt, P., Ciaian, P. and Gomez y Paloma, S. 
(2018). Corporate R&D and the performance of food-processing firms: Evidence 
from Europe, Japan and North America. Bio-based and Applied Economics 7(3), 
233-247.

ISTAT (2014). Symbola, Secondo Rapporto sulla Meccanizzazione Agricola Verde.
Klerkx, L., Hall, A. and Leeuwis, C. (2009). Strengthening agricultural innovation capac-

ity: are innovation brokers the answer? International Journal of Agricultural Resourc-
es, Governance and Ecology 8(5-6): 409-438.

Koundouri, P., Nauges, C. and Tzouvelekas, V. (2006). Technology adoption under pro-
duction uncertainty: theory and application to irrigation technology. American Jour-
nal of Agricultural Economics 88(3): 657-670

Läpple, D., Renwick, A. and Thorne, F. (2015). Measuring and understanding the drivers 
of agricultural innovation: Evidence from Ireland. Food Policy 51: 1-8.

Levins, R.A. and Cochrane, W.W. (1996). The treadmill revisited. Land Economics 72(4), 
550-553.

Lin, J.Y. (1991). Education and innovation adoption in agriculture: evidence from hybrid 
rice in China. American Journal of Agricultural Economics 73(3): 713-723

Lindner, R.K. (1987). Adoption and diffusion of technology: an overview. In ACIAR pro-
ceedings series.

Marra, M., Pannell, D.J. and Ghadim, A.A. (2003). The economics of risk, uncertainty 
and learning in the adoption of new agricultural technologies: where are we on the 
learning curve? Agricultural Systems 75(2): 215-234

Mairesse, J. and Mohnen, P. (2010). Using innovation surveys for econometric analysis. In 
Hall, B.H., and Rosenberg, N. (Eds.), Handbook of the Economics of Innovation (Vol. 
2). Elsevier, 1129-1155.

Nelson, R.R. and Winter, G.S. (1982). An evolutionary theory of economic change. Cam-
bridge: Harvard U. Press.

Ramos-Sandoval, R., Más Verdú, F. and García Álvarez-Coque, J.M. (2018). Do research 
and extension services improve small farmers’ perceived performance?. New Medit 
4: 7-19.

Ramos-Sandoval, R., García Álvarez-Coque, J.M. and Más-Verdú, F. (2019). Innovative 
capabilities of users of agricultural R&D services. Regional Science Policy & Practice 
11(2), 295-305.

Reimers, M. and Klasen, S. (2013). Revisiting the role of education for agricultural pro-
ductivity. American Journal of Agricultural Economics 95(1), 131-152.



208 Michele Vollaro, Meri Raggi, Davide Viaggi

Rezaei-Moghaddam, K. and Salehi, S. (2010). Agricultural specialists intention toward 
precision agriculture technologies: integrating innovation characteristics to technol-
ogy acceptance model. African Journal of Agricultural Research 5(11): 1191-1199.

Rogers, E.M. (1983). Diffusion of innovations. Simon and Schuster.
Röling, N. (1990). The agricultural research-technology transfer interface: a knowledge 

systems perspective. In Kaimowitz, D. (Eds.), Making the link: Agricultural research 
and technology transfer in developing countries. Westview Press Boulder, San Fran-
cisco & London.

Röling, N.G. (1994). Agricultural knowledge and information systems. In Blackburn, D.J. 
(Eds.), Extension handbook: Processes and practices. Toronto, Canada: Thompson 
Educational Publ. Inc.

Ruttan, V.W. (1996). What Happened to Technology Adoption‐Diffusion Research? Socio-
logia Ruralis 36(1): 51-73

Sauer, J. and Vrolijk, H. (2019). Innovation and performance–evidence at micro level. 
Applied Economics 51(43): 4673-4699.

SCAR, E.U. (2012). Agricultural knowledge and innovation systems in transition–a reflec-
tion paper. Brussels: European Commission, Standing Committee on Agricultural 
Research-Collaborative Working Group on Agricultural Knowledge and Innovation 
System (CWG AKIS).

Sunding, D. and Zilberman, D. (2001). The agricultural innovation process: research and 
technology adoption in a changing agricultural sector. Handbook of Agricultural 
Economics 1: 207-261

Thirtle, C.G. (1985). Induced Innovation in United States Field Crops, 1939-78. Journal of 
Agricultural Economics 36(1): 1-14.

Walder, P., Sinabell, F., Unterlass, F., Niedermayr, A., Fulgeanu, D., Kapfer, M., Melcher, 
M.and Kantelhardt, J. (2019). Exploring the Relationship between Farmers’ Innova-
tiveness and Their Values and Aims. Sustainability 11(20): 5571.

Walker, T., Ryan, J. and Kelley, T. (2010). Impact assessment of policy-oriented interna-
tional agricultural research: Evidence and insights from case studies. World Develop-
ment 38(10): 1453-1461.

Annex

Obs Mean Std. 
Dev. Min Max

Structural data
Zootechnics specialisation 300 0.08 0.26 0 1
Fruit specialisation, including grape and olives 300 0.26 0.44 0 1
Cereal specialisation 300 0.40 0.49 0 1
Protein crop specialisation 300 0.06 0.23 0 1
Arable crop specialisation, including horticultural crops 300 0.62 0.49 0 1
Presence of ancillary activity: yes=1; no=0 300 0.26 0.44 0 1



209Innovation adoption and farm profitability: what role for research and information sources?

Obs Mean Std. 
Dev. Min Max

Sale contracts 300 0.33 0.47 0 1
Share of rented land over total land 300 0.20 0.30 0 1
Own land 300 17.03 27.82 0 300
Rented land 300 9.72 25.11 0 200
Total Land 300 26.75 45.67 0 500
Number of tractors 300 3.43 2.83 0 20
Number of operational machines 300 3.15 2.32 0 9

Demographic data
Individual farm: yes=1; no=0 300 0.80 0.40 0 1
Family farm: yes=1; no=0 300 0.96 0.20 0 1
Family labour 285 1.89 1.09 0 7
Family labour Full Time 285 1.35 0.89 0 6
Family labour Part Time 285 0.54 0.86 0 4
Education inferior than medium school =1 300 0.22 0.41 0 1
Education superior than elementary school =1 300 0.73 0.45 0 1
Education superior than high schoo l=1 300 0.09 0.29 0 1
Specialized Ag education =1 300 0.46 0.50 0 1
Family income from Ag <30% =1 142 0.55 0.50 0 1
Family income from Ag <50% =1 176 0.69 0.46 0 1

Considerations
Important innovations in last 20 years: yes=1; no=0 300 0.47 0.50 0 1
Continue farming in 5 years:  
yes=3; maybe yes=2; maybe no=1; no=0 277 2.36 0.79 0 3

Introduce innovation in next 5 years: yes=1; no=0 176 0.35 0.48 0 1
Innovation important for competitiveness:  
not at all=0; little=1; enough=2; much=3 272 2.42 0.72 0 3

CAP help innovation adoption:  
not at all=0; little=1; enough=2; much=3 248 1.57 1.00 0 3

CAP necessary for supporting agriculture:  
not at all=0; little=1; enough=2; much=3 267 2.19 0.97 0 3

Description of data for non-innovators (reasons for not innovating)
No introduction = 1 300 0.60 0.49 0 1
No introduction for high costs = 1 179 0.47 0.50 0 1
No introduction for ethical reasons = 1 179 0.01 0.11 0 1
No introduction for too bureaucracy = 1 179 0.05 0.22 0 1
No introduction for high risks = 1 179 0.06 0.23 0 1
No introduction for quitting activity soon = 1 179 0.09 0.29 0 1
No introduction for negative past experiences = 1 179 0.02 0.13 0 1
No introduction for keeping traditions = 1 179 0.10 0.30 0 1
No introduction for other reasons = 1 179 0.28 0.45 0 1

Description of data for the subsample of innovators
Number of introduced innovations 300 0.71 1.16 0 8
Introduction of innovation: yes=1; no=0 300 0.40 0.49 0 1



210 Michele Vollaro, Meri Raggi, Davide Viaggi

Obs Mean Std. 
Dev. Min Max

Year of introduction of the innovation 109 2007 6.01 1995 2015
Age of innovation wrt to introduction 109 8.03 6.01 0 20
Intro for reducing risks = 1 121 0.11 0.31 0 1
Intro for diversifying ag activity = 1 121 0.14 0.35 0 1
Intro for reducing costs = 1 121 0.55 0.50 0 1
Intro for increasing production = 1 121 0.46 0.50 0 1
Other reasons (most increasing profitability and reducing labour) 121 0.27 0.45 0 1
Reaction to increase in input prices 121 0.49 0.50 0 1
Reaction to reduction in output prices 121 0.52 0.50 0 1
Anticipate inputs markets trend 121 0.36 0.48 0 1
Anticipate outputs markets trend 121 0.37 0.49 0 1
External help from private or seller 120 0.37 0.48 0 1
External help from public institutions 120 0.01 0.09 0 1
No external financial support for introducing innovation 120 0.56 0.50 0 1
Level of self-financing: 0=less than 5.000; 3=more than 50.000 91 1.85 1.10 0 3

Type of innovations
Biological and Genetic innovations 121 0.07 0.25 0 1
Agronomical and Zoological innovations 121 0.06 0.23 0 1
Mechanical innovations 121 0.37 0.49 0 1
Informatics innovations 121 0.02 0.13 0 1
Energy and water saving innovations 121 0.22 0.42 0 1
Diversification innovation 121 0.12 0.33 0 1
Market strategies innovations 121 0.02 0.13 0 1

Information about origin of innovation
Source of information about innovation: 
external=1; self produced=0 121 0.69 0.47 0 1

Knowledge of innovation origin from research 121 0.44 0.50 0 1

Effects of introduced innovation on economic performance
All effects: presence of (positive) effect=1; otherwise=0 121 0.87 0.34 0 1
Cost reduction in % 63 17.81 20.94 0 90
Cost reduction: yes=1; no=0 63 0.71 0.46 0 1
Production increment in % 71 16.17 24.02 0 100
Production increment: yes=1; no=0 71 0.65 0.48 0 1
Value added increment in % 75 11.20 18.56 0 100
Value added increment: yes=1; no=0 75 0.52 0.50 0 1
Quality increment >0:  
very high, high and low=1; nothing=0 121 0.80 0.40 0 1

Quality increment >1:  
very high, high=1; otherwise=0 121 0.60 0.49 0 1

Quality increment:  
not at all=0; little=1; enough=2; much=3 110 1.64 1.04 0 3


	Consumers’ rationality and home-grown values for healthy and environmentally sustainable food
	Simone Cerroni1,2,3,*, Verity Watson4, Jennie I. Macdiarmid5
	The wellbeing of smallholder coffee farmers in the Mount Elgon region: a quantitative analysis of a rural community in Eastern Uganda
	Anna Lina Bartl
	Agricultural Sector Performance, Institutional Framework and Food Security in Nigeria
	Romanus Osabohien1,3,*, Evans Osabuohien1,3, Precious Ohalete2,3
	Innovation adoption and farm profitability: what role for research and information sources?
	Michele Vollaro1,*, Meri Raggi2, Davide Viaggi1
	Determinants of Farm Households’ Willingness to Accept (WTA) Compensation for Conservation Technologies in Northern Ghana 
	Evelyn Delali Ahiale1,*, Kelvin Balcombe2, Chittur Srinivasan2