Journal of Marketing Analytics

https://doi.org/10.1057/s41270-021-00137-4

ORIGINAL ARTICLE

Producers’ brand‑dealer dual loyalty to capital equipment

Pedro Harmath1 · Roberto Feeney2 · Josefa Ramoni‑Perazzi3 

Revised: 9 August 2021 / Accepted: 20 September 2021

© The Author(s), under exclusive licence to Springer Nature Limited 2021

Abstract
This paper aims to identify the factors that affect agricultural producers’ loyalty to brand and dealer, jointly and individually,
when buying agricultural capital equipment. A second objective is exploring the influence of one type of loyalty over the
other, for which we resort to different settings. Our findings show distinctive determining factors for each type of loyalty, the
bi-univocity of these two loyalties, and the necessary farm profile to achieve dual loyalty. Agricultural equipment companies
demand this information to target farmers and design marketing strategies to make farmers loyal or disloyal, strengthen their
distribution channels, and identify the best communication lines to reach potential and current customers.

Keywords  Agricultural capital equipment · Loyalty to brand and dealer · Argentinian Ag equipment market · Probit
regression models

Introduction Alternatively, producers can be loyal both to the brand and

Agricultural capital equipment markets are fiercely competi-
tive (Reportlinker 2019), and firms selling these products
need agricultural producers’ loyalty to be profitable (Ag
Equipment Intelligence 2018). In this purchasing process,
establishing long-term bonds with producers can become
a key asset not only for manufacturers but also for dealers.
Dealers are important figures to capital equipment manu-
facturers as they sell new products and provide after-sales
services: fast delivery of parts and services in the event the
equipment fails or collapses (Walley et al. 2007).
There is often interplay between brand and dealer loyalty,
as many times brand loyalty may depend on the dealer’s
performance (Verhoef et al. 2007). In this context, producers
can show their loyalty to the Ag capital equipment brand or
the dealer, according to the specific value each can supply.
the dealers at the same time.
However, the factors that explain brand and dealer loyalty
for capital equipment, and the interplay between brand and
dealer loyalty, have scarcely been studied in the agribusiness
context. Thus, this research aims to fill a gap in the literature
concerning the issue of which are the determining factors of
agricultural producers’ loyalty to capital equipment brands
and dealers, what is the influence of one type of loyalty over
the other, and which are the factors that define producers’
dual loyalty to agricultural capital equipment.
Consequently, this paper has three main goals: (1) iden-
tify the factors that determine Argentinian producers’ brand
loyalty and dealer loyalty when in need of agricultural capi-
tal equipment, using univariate probit models; (2) study the
influence of one type of loyalty over the other (brand and
dealer loyalty), using seemingly unrelated bivariate probit
models; and (3) identify the factors that determine dual loy-
alty, using multinomial probit models.

* Josefa Ramoni‑Perazzi
jramonip@uis.edu.co Literature review
1
Department of Mathematics, Austral University, Rosario,
Argentina Agricultural producers act as business customers when inter-
2
Center for Food & Agribusiness, Austral University, Rosario,
acting with capital equipment suppliers, just like in a non-
Argentina farm business-to-business (B2B) environment. Purchasers
3
Escuela de Economía y Administración, Universidad
in the B2B markets are characterized by being professional,
Industrial de Santander, Carrera 27, Calle 9, rational, and expert buyers (Harbor 2006).
Bucaramanga 680001, Colombia

Vol.:(0123456789)

Despite the significant contribution of brands to indus-
trial firms’ value and profitability (Cassia and Magno 2019;
Nyström et al. 2018; Cassia et al. 2017; Viardot 2017; Wal-
ley et al., 2007), branding in industrial contexts is an under-
researched area due to the implicit assumption that brands
are not relevant for industrial customers (Seyedghorban et al.
2016; Hinterhuber and Hinterhuber 2012). However, previ-
ous research has established the importance of reputation
and trust that brands may convey to industrial customers
(Lynch and De Chernatony 2004), and the fact that brands
can mitigate the uncertainty and risk that professional pur-
chasers experience in industrial markets (Leek and Christo-
doulides 2012; Mudambi 2002).
Industrial brands are valuable for end-users and industrial
distributors or dealers. Manufacturers’ brands are an impor-
tant factor in the dealership´s performance, as the brand can
be a valuable resource to attract customers, increase demand,
revenues, and profits. At the same time, the dealer is the
customer’s point of contact with the product; higher levels of
service quality tend to increase customer satisfaction. This
would result in loyalty to the dealer, which in turn would
lead to greater loyalty to the brand (Anisimova and Mavondo
2014; Glynn 2010; Huber and Herrmann 2001).
Focusing on the relationship between brands and dealers,
Hinterhuber and Hinterhuber (2012) analyzed the impor-
tance of the manufacturer’s brands for industrial dealers in
the agrochemical industry. The analysis was based on a set
of three unrelated studies, one of which includes the role
of the customer (end-users). This last study was conducted
through face-to-face interviews with farmers who were
asked to rank the herbicides attributes they value the most.
Factors related to quality came first, followed by the brand
in second place.
According to Hinterhuber and Hinterhuber (2012), brands
drive industrial retailers’ and users’ decisions more than any
change in price and margins, implying that brand loyalty
comes first and has an impact on dealer loyalty. The brand
itself denotes a perception of quality, credibility, and reli-
ability that leads to relatively higher returns; price, then,
becomes a not so relevant variable. Therefore, industries
should focus on building a brand rather than on designing
marketing strategies based on prices, which calls for a coop-
erative relationship between manufacturers and retailers or
dealers. Trust in a brand facilitates the production process,
leading to deep-rooted brand loyalty.
Similarly, Walley et al. (2007) analyzed the role of brand-
ing in the purchase decisions made in the UK tractor market,
based on the information collected from a sample of 428
farmers and contractors. Participants were asked to rank the
attributes that influence their decisions. The results of the
conjoint analysis pointed out that brand name was the most
important factor, followed by price, and service (dealer prox-
imity and quality of service).
P. Harmath et al.
According to Huber and Herrmann (2001), custom-
ers do not see brand and dealer loyalty as one unit. The
authors used confirmatory factor analysis with unweighted
least squares to analyze the information collected through
telephone interviews, from a sample of 615 driver license
holders who had purchased a vehicle in Germany. They
found evidence that supports the idea that customer satis-
faction with the product increases brand loyalty, which, in
turn, increases dealer loyalty. At the same time, satisfaction
with the services provided by the dealer can increase dealer
loyalty and lead to greater brand loyalty.
In the same line of thought, Akaeze and Akaeze (2017)
concluded that customer satisfaction, product quality, and
service quality are, in that order, the most important factors
affecting customer loyalty to the dealer in the automobile
market. The authors arrived at this conclusion by using the
information gathered from a structured questionnaire, self-
administered to 50 customers in New York, and analyzed
it in the theory of reasoned action conceptual framework.
Verhoef et  al. (2007) also established that brand and
dealer loyalty are interconnected. Consumers’ decisions
are influenced by the quality of the service provided by
the dealer and, at the same time, dealer loyalty depends on
brand loyalty. The lower the brand tier, the more oppor-
tunities for dealers to add value to the brand and improve
customer satisfaction. Once again, the purchasing decision
starts with the brand and moves to the dealer, meaning that
brand loyalty precedes dealer loyalty. At the same time,
dealer performance affects brand retention. The authors
focused on disloyalty (switching behavior) rather than loy-
alty and accounted for brand tier since it determines the level
at which dealers may add value and develop customer loy-
alty. The study also considered information regarding the
alignment between brand and dealer, and the institutional
environment.
Ewing (2000) found evidence supporting the idea that
the expectation of a customer purchasing a given brand in
the future is influenced by past brand or dealer repeated pur-
chasing behavior. However, past repeated brand or dealer
purchasing behavior does not seem to affect anyone’s expec-
tation to purchase from a given dealer. The author concludes
that the brand/end-user interface offers a greater predictive
capacity than the retail/end-user interface. The hypotheses
were tested with a one-way analysis of variance, using infor-
mation from a self-administered questionnaire answered by
a random sample of small and medium-sized new-vehicle
owners in Johannesburg.
In all these papers, loyalty to brand and dealer are pre-
sented as related concepts, as each influences the other.
However, brand loyalty impacts first and foremost on dealer
loyalty, since product satisfaction leads to brand loyalty
and only indirectly would impact on dealer loyalty. Subse-
quently, the dealer loyalty contributes to loyalty to brand:
Producers’ brand‑dealer dual loyalty to capital equipment

a customer satisfied with the dealers’ services would have Methods

a positive effect on dealer loyalty, which would lead to
higher brand loyalty, in a sort of loop effect (Anisimova and According to Cameron and Trivedi (2005), Greene (2012),
Mavondo 2014; Glynn 2010; Harbor 2006; Huber and Her- Seyoum (2018), and Wooldridge (2010), among others, let
rmann 2001). Y ∗ be a latent variable such as the producers’ commitment
The same business and economic principles of B2B for brands or dealers if brand loyalty or dealer loyalty is
branding should hold for agricultural capital equipment mar- being modeled. Then, the natural regression model for this
kets (Walley et al. 2007; Harbor 2006; Kool et al. 1997). variable is given by
However, the authors are ignorant of the factors that lead to
brand and dealer loyalty, which directed them to questions Y ∗ = X𝛽 + 𝜀 (1)
asking about the factors that explain brand loyalty and dealer where 𝜀 is the disturbance component or error assumed to
loyalty: Are they the same, or do brand and dealer loyalty be normal with mean zero and variance one, (𝜀|X) ∼ N[0,1] .
involve different sets of variables? On the other hand, are However, model (1) cannot be estimated as Y ∗ is not
there factors that can explain loyalty to brand and dealer? observed. Instead, we observe
{
1 if Y ∗ > 0,
Y= (2)
Methodology 0 if Y ∗ < 0

Data The univariate probit model can be written as

�
( � ) X 𝛽

∫
This paper´s main source of information is the survey “The P(Y = 1|X) = Φ X 𝛽 = 𝜙(z)dz (3)
Needs of the Argentine Agricultural Producer” conducted −∞
during June and July 2017. The survey aims at analyzing
where Φ(⋅) is the standard normal cumulative distribu-
farmers’ purchasing behavior and at understanding their
tion function (cdf), and the correspondent standard nor-
underlying preferences when making decisions. It defines
mal probability
√ density function (pdf) is the derivative
agricultural producers as physical or legal persons growing
𝜙(z) = (1∕ 2𝜋)exp(−z2 ∕2) . The coefficients (𝛽) are esti-
soybeans, wheat, corn, and other grains on owned and/or
mated using the maximum likelihood estimation.
rented land. (for more details, see e.g. Mohammadi et al.
This study also considers the application of a seemingly
2020). In this research, we consider a random sample of
unrelated bivariate probit model, used when two equations
564 farmers, where ‘loyalty to brands’, ‘loyalty to dealers’,
are estimated and the dependent variable on one of them is
‘dual loyalty’ or ‘concurrent loyalty to brands and dealers’
independent in the other, assuming the possibility that one
are defined as dependent variables.1
kind of loyalty influences the other. Let Y1∗ be the producers’
Based on the general literature, we selected the independ-
commitment for dealers (brands), and Y2∗ be the producers’
ent variables that influence these types of loyalty. The lit-
commitment for brands (dealers). The resulting equations
erature suggests that certain demographic characteristics of
are:
the producer such as age, education, and income, as well as
other characteristics related to the nature of the firm such Y1∗ = X1 𝛽1 + 𝜀1 (4)
as farm size, can influence loyalty to brands and dealers. It
is expected that price, performance, the valuation of possi-
Y2∗ = 𝛾Y 1 +X 2 𝛽2 + 𝜀2 (5)
ble differences between products, and preferences towards
different means of communication that producers use to where 𝜀1 and 𝜀2 are jointly normal with zero means, unit
acquire capital equipment provided by the supplier, play an variances, and correlation 𝜌 . Succinctly
important role in determining brand loyalty. Additionally, we ( ) [( ) ( )]
include variables that are relevant in the Argentine context 𝜀1 0 1 𝜌
|X1 , X2 ∼ N , (6)
and which have been acknowledged by the literature: the 𝜀2 0 𝜌1
distance between the place where the producer often lives
and his/her farm, the percentage of rented land, and produc- with
ers’ cognitive styles (in terms of being analytical, intuitive, {
1 if Y1∗ > 0,
or balanced). The list of the variables is shown in Table 1. Y1 =
0 if Y1∗ < 0 (7)

and
1
  These were built through cluster analysis. For details, see Appendix
A.
 P. Harmath et al.

Table 1  Dependent and independent variables list

Name of the variable Categories of response

Dependent variable
 BLOYALTY Loyalty to brands 1 = loyal; 0 = disloyal
 DLOYALTY Loyalty to dealers 1 = loyal; 0 = disloyal
 BDLOYALTY1 Dual loyalty (loyalty to brands and dealers 1 = loyal to both brands and dealers; 0 = oth-
simultaneously) erwise
 BDLOYALTY2 1 = loyal to brands and disloyal to dealers
0 = otherwise
 BDLOYALTY3 1 = loyal to dealers and disloyal to brands
0 = otherwise
 BDLOYALTY4** 1 = disloyal to both brands and dealers;
0 = otherwise
Independent variable
 Dimension 1 for brands and dealers: farm and farmer characteristics
  EDUC Farmer education level 0 = high school or less
1 = more than high school
  AGEG1* Farmer age group 1 =  ≤ 44; 0 = otherwise
  AGEG2 1 = 45–54; 0 = otherwise
  AGEG3 1 =  ≥ 55; 0 = otherwise
  RESID Residence (distance to farm) 0 = In the facilities or less than 50 kms. away
1 = more than 50 kms. away
  FSIZE1* Farm size 1 = mid-sized; 0 = otherwise
  FSIZE2 1 = commercial; 0 = otherwise
  FSIZE3 1 = large; 0 = otherwise
  PRL % of rented land 0 = less than 50%
1 = 50% or more
  GS1* Gross Sales (U$D) 1 =  ≤ 499,999; 0 = otherwise
  GS2 1 = 500,000–999,999
0 = otherwise
  GS3 1 =  ≥ 1,000,000; 0 = otherwise
 Dimension 2 for brands: farmer beliefs and attitudes
  CS Cognitive Style 1 = analytic
0 = intuitive/balanced
  DBEBP1* Differences between expendable and branded 1 = strongly disagree—disagree; 0 = otherwise
  DBEBP2 products 1 = partially disagree
0 = otherwise
  DBEBP3 1 = partially agree—agree—strongly agree;
0 = otherwise
 Dimension 3 for brands: products characteristics
  PRICE Appreciation of the attribute: Price 1 = most important
0 = least important/neutral
  PERFORM Appreciation of the attribute: Performance 1 = most important
0 = least important/neutral
 Dimension 4 for brands: media exposure
  MEDSOU More visited media source 1 = traditional and personal communication;
0 = social media
  MEANVAL Index measuring reported media exposure
ranging from 1 to 6
 Dimension 2 for dealers: supplier characteristics
  FPDR Financing provided by dealer/retailer 0 = 0–50%; 1 = 51–100%
 Dimension 3 for dealers: relationship with supplier
  REL Appreciation of the attribute: Relationship with 1 = most important
dealer 0 = least important/neutral

*Category of reference or control, **base outcome

Producers’ brand‑dealer dual loyalty to capital equipment

{ � � � � � � �
�
1 if Y2∗ > 0, 𝜕Φ2 X1 𝛽1 , X2 𝛽2 , 𝜌 � � � X2 𝛽2 − 𝜌X1 𝛽1
Y2 =
0 if Y2∗ < 0 (8) = 𝜙 Xi 𝛽i Φ √ 𝛽i , i = 1,2
𝜕Xi 1 − 𝜌2
respectively. In short, the seemingly unrelated probit (15)
( � )
model can be denoted by 𝜕Φ X 𝛽i ( � )
( ) ( � ) = 𝜙 X 𝛽i 𝛽j , i = 0,1, 2,3 (16)
� 𝜕Xj
P Y1 = 1, Y2 = 1|X1 , X2 = Φ2 X1 𝛽1 , 𝛾Y1 , X2 𝛽2 , 𝜌 (9)

where Φ2 (⋅) is the bivariate standard normal cdf with our

correspondent bivariate standard normal pdf 𝜙2 (⋅) . The
development of maximum likelihood of the expression (9) is
( ) Results and discussion
appropriate for the estimation of the coefficients 𝛽1 , 𝛾, 𝛽2 , 𝜌  ,
respectively.
Descriptive statistics and chi‑square test
On the other hand, regarding the Argentinian producers’
of association
dual loyalty, let Y1∗ be the producers’ commitment for both
brands and dealers, Y2∗ be the producers’ commitment for
The results shown in Table 2 indicate that farmers are more
brands but not for dealers, Y3∗ be the producers’ commitment
likely to be loyal to the dealer rather than to the brand. In any
for dealers but not for brands, and Y0∗ be the general indif-
case, loyal farmers are never the majority. Producers typi-
ference or producers’ lack of commitment for both brands
cally have more than a high school diploma, especially those
and dealers. In this case, the regression equation is given by
who consider themselves brand loyal. Their average age is
Yi∗ = X𝛽i + 𝜀i (10) 47 years, older among disloyal. They tend to live close to the
farm and have gross sales that do not exceed half a million
for i = 0,1, 2,3 , where the errors are jointly normally dis- dollars, which may be explained by the fact that 60% or more
tributed, with are commercial farms. They mostly rent less than 50% of the
total area, especially those that are not loyal to either brand
(𝜀|X) ∼ N[0, Σ] (11)
or dealer. Except for brand loyalists, farmers typically use an
where 𝜀 = [𝜀0 , 𝜀1 , 𝜀2 , 𝜀3 ] , such that Σ is the correlation intuitive style when making purchasing decisions. Also, they
matrix between these random components with values one tend to resort to traditional media sources to find information
at the principal diagonal of Σ , and about the equipment they are planning to buy, and less than
{ one-third of the producers are financed by dealers for more
1 if Yi∗ > 0, than 50% of their purchases.
Yi =
0 if Yi∗ < 0 (12)
On the other hand, before adjusting the corresponding
probit models, as a first step, we review the relationship
The multinomial probit model can be written as
between each dependent variable and the corresponding
( ) ( � )
�
X 𝛽i explanatory variables, making use of a Chi-square test.
∫
P Yi = 1|X = Φ X 𝛽i = 𝜙(z)dz (13) As shown in Table 3, in the brand loyalty column (BLOY-
−∞ ALTY), all the explanatory variables are associated with
where Φ(⋅) and 𝜙(⋅) are the respective standard normal cdf the Chi-square test, except the gross sales volume (GS), and
and pdf like in the model defined in Eq. (3), considering one the appreciation of the attribute price (PRICE). Regarding
of the categories in the dependent variable Yi as the base dealer loyalty (DLOYALTY), all the explanatory variables
outcome, with i = 0,1, 2,3. Naturally, the coefficients ( 𝛽i , all are associated with this loyalty, except the producer's educa-
the elements of Σ ) can be estimated using the maximum tion level (EDUC), the farm size (FSIZE), and the apprecia-
likelihood estimation. tion of the producer-dealer (REL) relationship.
Finally, the marginal effects for the univariate probit model For dual loyalty (BDLOYALTY), all the explanatory
(3), the seemingly unrelated probit model (9), and the multi- variables are associated with this loyalty except the pro-
variate probit model (13) are defined by ducer education level (EDUC), the distance between the
( � ) place where the producer usually resides and his/her farm
𝜕Φ X 𝛽 ( � ) (RESID), the appreciation of the price attribute (PRICE),
= 𝜙 X 𝛽 𝛽j (14) and the assessment of the relationship between the producer
𝜕Xj
and the dealer (REL). Regarding the media assessment
(MEANVAL), the only quantitative variable in our study, it
differs significantly on average between producers that are
loyal or disloyal to agricultural machinery brands; but not
 P. Harmath et al.

Table 2  Descriptive statistics Variable Description Loyalty to Loyalty to dealers

brands
Loyal Disloyal Loyal Disloyal

EDUC Level of education (%)

Highschool (0) 38.03 27.07 32.30 30.29
More (1) 61.97 72.93 67.70 69.71
RESID Distance to farm (%)
 ≤ 50 Kms (0) 69.95 64.39 68.87 64.50
 > 50 Kms (1) 30.05 35.61 31.13 35.50
FSIZE Farm size (%)
Mid (1) 63.38 64.10 59.14 67.75
Commercial (2) 29.11 21.37 29.57 19.87
Large (3) 7.51 14.53 11.28 12.38
PRL Rented land (%)
 ≤ 50% (0) 56.81 67.52 59.14 67.10
 > 50% (1) 43.19 32.48 40.86 32.90
GS Gross Sales (%)
 < 500,000 US$ (1) 45.54 42.74 42.80 44.63
500,000–999,999 US$ (2) 24.88 31.05 26.85 30.29
 > 999,999 (3) 29.58 26.21 30.35 25.08
CS Cognitive Style (%)
Intuitive (0) 45.54 55.27 51.75 51.47
Analytical (1) 54.46 44.73 48.25 48.53
DBEBP Differences between expendable and
branded products (%)
Strongly disagree/disagree (1) 30.99 59.26 42.80 53.42
Partly disagree (2) 16.43 7.98 14.79 8.14
Agree/strongly agree (3) 52.58 32.76 42.41 38.44
PRICE Importance of attribute price (%)
Less importance/neutral (0) 71.83 77.78 75.88 75.24
Most important (1) 28.17 22.22 24.12 24.76
PERFORM Importance of attribute performance (%)
Less importance/neutral (0) 50.70 43.87 46.69 46.25
Most important (1) 49.30 56.13 53.31 53.75
REL Importance of relation with dealer (%)
Less importance/neutral (0) 78.40 78.92 78.21 79.15
Most important (1) 21.60 21.08 21.79 20.85
MEDSOU Most frequent media source (%)
Traditional (1) 61.50 54.99 57.20 57.65
Personal (2) 20.66 12.25 15.18 15.64
Social (3) 17.84 32.76 27.63 26.71
FPDR Financing provided by dealer/retailer (%)
 ≤ 50% (0) 69.01 80.91 72.37 79.80
 > 50% (1) 30.99 19.09 27.63 20.20
AGE Age of farmers (years)
Mean 43.38 49.49 45.74 48.39
Standard deviation 12.13 11.62 11.89 12.29
MEANVAL Index reported media exposure (1–6)
Mean 3.47 3.19 3.35 3.25
Standard deviation 0.69 0.78 0.71 0.77
Number of observations 213 351 257 307
Percent of total (%) 37.77 62.23 45.57 54.43
Producers’ brand‑dealer dual loyalty to capital equipment

Table 3  Chi-square test between the dependents and each independent variable

Variable Name of the variable BLOYALTY: loy- DLOYALTY: loy- BDLOY-
alty to brands alty to dealers ALTY: dual
loyalty

EDUC Farmer education level 3.936** 1.963 5.279

AGEG Farmer age group 59.016*** 17.384*** 60.330***
RESID Residence (distance to farm) 2.942* 2.702* 5.335
FSIZE Farm size 9.304*** 2.997 14.492**
PRL % of rented land 16.212*** 18.869*** 25.492***
GS Gross Sales (U$D) 4.078 8.609** 21.861***
CS Cognitive Style 23.512*** 25.010***
DBEBP Differences between expendable and branded products 52.026*** 53.391***
PRICE Appreciation of the attribute: Price 1.343 5.235
PERFORM Appreciation of the attribute: Performance 7.882*** 11.559***
MEDSOU More visited media source 9.697*** 12.280***
MEANVAL Index measuring reported media exposure ranging from 1 to 6 tc = 3.414*** Fc = 0.909
FPDR Financing provided by dealer 6.163** 9.781**
REL Appreciation of the attribute: Relationship with dealer 1.341 5.561

T—mean sample comparison test with equal variances for MEANVAL. Levene’s test for equality of variances was not significant, with
Fc = 2.277. Chi-square statistic between BLOYALTY and DLOYALTY = 115.4***
(*), (**) and (***) indicate significance at the 0.10, 0.05 and 0.01, respectively

so when the four categories of loyalty to brands and deal-
ers together are considered. As an additional test, the Chi-
square test between BLOYALTY and DLOYALTY shows
that both variables are not statically independent, which is
why it makes sense to examine the influence of one type of
loyalty over the other, as presented below (see note at the
end of Table 3).
Univariate analysis for brand loyalty and dealer
loyalty
To identify the factors affecting one type of loyalty or the
other, several univariate probit models were adjusted consid-
ering the variables identified in the previous section. Among
the results shown in Table 4, for analysis purposes, the focus
was on models (1.4) and (2.3) because they are the most
parsimonious (see note at the end of Table 4). For brand
loyalty (BLOYALTY), the factors that statistically determine
this variable are CS (+), DBEBP2 (+), DBEBP3 (+), PER-
FORM (+), and EDUC (−), AGEG2 (−), AGEG3 (−). As
to dealer loyalty (DLOYALTY), the factors that statistically
determine it are PRL (+), GS2 (+), GS3 (+), and AGEG2
(−), AGEG3 (−), RESID (−).
The average marginal effects of the statistically significant
coefficients indicate that analytical agricultural producers
are 8.88% more likely to be considered loyal to agricultural
machinery brands than intuitive or balanced producers.
Those who partially or strongly consider that there are differ-
ences between generic and trademark products are 19.59%
and 13.18%, respectively, more prone to be considered loyal
than those who do not consider such differences. Produc-
ers who attach great importance to product performance are
13.89% more likely to be considered loyal to such brands,
compared to those that give less value to this attribute. On
the other hand, agricultural producers with a higher level
of education are 9.50% less likely to be considered loyal to
agricultural machinery brands. Older producers (45–54, ≥
54 years old) are 16.8% and 28.5% less likely to be loyal to
agricultural machinery brands, respectively, when compared
to the younger ( ≤ years old).
Producers who rent at least 50% of the land where they
carry out their agricultural operations are 14.87% more
likely to be loyal to agricultural machinery dealers, com-
pared to those who rent less than 50%. Those with medium
and high gross sales volumes have 13.4% and 11.07% more
chances of loyalty than those with the lowest gross sales
volumes. As a counterpart, similar to observations made
about brand loyalty, older producers above 55 years old are
11.5% and 13.4% less likely to be loyal to distributors of
agricultural machinery compared with the youngers, below
55 years, while the loyalty of those who reside farther from
their farm is 11% less likely, compared to those who reside
closer or even in the same farm.
Bivariate analysis for brand loyalty and dealer
loyalty
To quantify the impact of one type of loyalty over the other,
some seemingly unrelated bivariate probit models were
adjusted. Table 5 shows the models where dealer loyalty
 P. Harmath et al.

Table 4  Univariate probit models for brand loyalty and dealer loyalty

Variable Dep. Variable: BLOYALTY Dep. Variable: LOYALTY
(1.1) (1.2) (1.3) (1.4) Marg. eff (2.1) (2.2) (2.3) Marg. eff

CONSTANT − 0.7136* − 0.6858* − 0.513*** − 0.539*** 0.1863 0.0275 0.0467

(0.3834) (0.3633) (0.1850) (0.1841) (0.1564) (0.1237) (0.1216)
EDUC − 0.272** − 0.275** − 0.2885** − 0.3201** − 0.095 − 0.1769
(0.1381) (0.1373) (0.1350) (0.1333) (0.1226)
AGEG2 − 0.53*** − 0.50*** − 0.529*** − 0.560*** − 0.168 − 0.2896** − 0.2990** − 0.3056** − 0.115
(0.1466) (0.1450) (0.1412) (0.1397) (0.1311) (0.1302) (0.1300)
AGEG3 − 0.90*** − 0.90*** − 0.958*** − 0.953*** − 0.285 − 0.380*** − 0.3437** − 0.3561** − 0.134
(0.1786) (0.1774) (0.1701) (0.1697) (0.1484) (0.1448) (0.1440)
RESID − 0.303** − 0.2731* − 0.2121 − 0.2720** − 0.2935** − 0.2940** − 0.110
(0.1419) (0.1399) (0.1318) (0.1254) (0.1222) (0.1222)
FSIZE2 0.0808 0.1788 − 0.0489
(0.1586) (0.1421) (0.1494)
FSIZE3 0.1152 0.3118 − 0.0285
(0.2383) (0.2004) (0.2236)
PRL 0.1026 0.0956 0.3483*** 0.3736*** 0.3960*** 0.1487
(0.1298) (0.1292) (0.1229) (0.1205) (0.1178)
GS2 0.0784 0.3717*** 0.3512*** 0.3570*** 0.1340
(0.1571) (0.1390) (0.1344) (0.1341)
GS3 0.2949 0.3228* 0.2805* 0.2948** 0.1107
(0.1966) (0.1807) (0.1469) (0.1460)
CS 0.309** 0.3108** 0.3284** 0.2971** 0.0888
(0.1328) (0.1324) (0.1292) (0.1275)
DBEBP2 0.64*** 0.624*** 0.650*** 0.655*** 0.1959
(0.2018) (0.2003) (0.1964) (0.1960)
DBEBP3 0.390** 0.3788** 0.435*** 0.440*** 0.1318
(0.1530) (0.1525) (0.1435) (0.1432)
PRICE − 0.0643
(0.1879)
PERFORM 0.391** 0.425*** 0.4650*** 0.4645*** 0.1389
(0.1610) (0.1241) (0.1215) (0.1213)
MEDSOU 0.2340 0.1961
(0.1532) (0.1501)
MEANVAL − 0.0200 − 0.0149
(0.0863) (0.0859)
FPDR 0.1154 0.1163
(0.1345) (0.1342)
REL − 0.1420
(0.1335)
Log-likelihood − 294.289 − 295.567 − 298.538 − 299.8417 − 367.6518 − 369.1304 − 369.5071
Likelihood Ratio Chi-squared 120.8*** 118.2*** 112.3*** 109.7*** 44.52*** 41.56*** 40.80***
Pseudo R2 0.1703 0.1667 0.1583 0.1546 0.0571 0.0533 0.0523
Correctly classified 70.74% 71.10% 71.28% 70.74% 62.94% 63.65% 63.30%

(1.1) included all the independent variables for brands; (1.2) included all the variables except GS, PRICE (see Chi-square test); (1.3) included
only the variables statistically significant in (1.1) and (1.2); (1.4) included only the variables statistically significant in (1.3)
(2.1) included all the independent variables for dealers; (2.2) included all the variables except EDUC, FSIZE, REL (see Chi-square test); (2.3)
included only the variables statistically significant in (2.1) and (2.2)
Standard errors in (). (*), (**) and (***) indicate significance at the 0.10, 0.05 and 0.01, respectively
Producers’ brand‑dealer dual loyalty to capital equipment

Table 5  Seemingly unrelated bivariate probit models: first dealer loyalty as the dependent, and then brand loyalty as the dependent
Dependent variable Variable (3.1.1) (3.2.1) (3.3.1) (3.4.1) Mar-
ginal  effects

DLOYALTY CONSTANT 0.1809 0.0260 0.0454 0.0573

(0.1569) (0.1234) (0.1211) (0.1221)
EDUC − 0.1769
(0.1228)
AGEG2 − 0.2859** − 0.2979** − 0.3067** − 0.3038** − 0.1206
(0.1314) (0.1302) (0.1298) (0.1296)
AGEG3 − 0.3790** − 0.3421** − 0.3548** − 0.3597** − 0.1427
(0.1485) (0.1449) (0.1440) (0.1444)
RESID − 0.2715** − 0.2959** − 0.2990** − 0.3237*** − 0.1285
(0.1258) (0.1224) (0.1225) (0.1220)
FSIZE2 − 0.0527
(0.1499)
FSIZE3 − 0.0354
(0.2233)
PRL 0.3516*** 0.3737*** 0.3926*** 0.3823*** 0.1503
(0.1232) (0.1205) (0.1183) (0.1198)
GS2 0.3757*** 0.3469** 0.3545*** 0.3450** 0.1354
(0.1398) (0.1366) (0.1346) (0.1357)
GS3 0.3226* 0.2864* 0.3130** 0.3230** 0.1268
(0.1804) (0.1472) (0.1486) (0.1440)
FPDR 0.1234 0.1209
(0.1326) (0.1340)
REL − 0.1381
(0.1343)
BLOYALTY CONSTANT − 1.7737*** − 1.7828*** − 1.5745*** − 1.7191***
(0.5895) (0.6128) (0.4224) (0.3451)
DLOYALTY 1.7635* 1.5535* 1.5929*** 1.8237*** 0.5280
(1.0686) (0.9049) (0.6222) (0.5262)
EDUC − 0.2028 − 0.2451* − 0.2472* − 0.2596* − 0.0861
(0.1885) (0.1478) (0.1454) (0.1435)
AGEG2 − 0.4504* − 0.4648** − 0.4664*** − 0.4448** − 0.1362
(0.2627) (0.2012) (0.1798) (0.1889)
AGEG3 − 0.7531** − 0.7917*** − 0.8109*** − 0.7311*** − 0.2054
(0.3597) (0.2666) (0.2474) (0.2587)
RESID − 0.2148 − 0.2233 − 0.1780
(0.2208) (0.1874) (0.1546)
FSIZE2 0.1249 0.1605
(0.1678) (0.1620)
FSIZE3 0.1407 0.2689
(0.2554) (0.2291)
PRL − 0.0796 − 0.0639
(0.2125) (0.1952)
GS2 − 0.0973
(0.2334)
GS3 0.1225
(0.2710)
CS 0.2999** 0.2955** 0.3021** 0.2617* 0.0844
(0.1446) (0.1433) (0.1414) (0.1347)
 P. Harmath et al.

Table 5  (continued)
Dependent variable Variable (3.1.1) (3.2.1) (3.3.1) (3.4.1) Mar-
ginal  effects

DBEBP2 0.5909*** 0.6110*** 0.6146*** 0.5950*** 0.2145

(0.2291) (0.2134) (0.2100) (0.2081)
DBEBP3 0.4036** 0.4055*** 0.4330*** 0.4185*** 0.1377
(0.1715) (0.1639) (0.1552) (0.1546)
PRICE − 0.1950
(0.2009)
PERFORM 0.3376 0.4765*** 0.5013*** 0.4872*** 0.1543
(0.2099) (0.1359) (0.1324) (0.1351)
MEDSOU 0.1198 0.0913
(0.1619) (0.1606)
MEANVAL 0.0355 0.0444
(0.0909) (0.0923)
rho − 0.3095 − 0.1674 − 0.1948 − 0.3649
(0.7745) (0.6051) (0.4234) (0.3951)
Log-likelihood − 614.1192 − 616.7137 − 618.5792 − 619.2541
Wald Chi-squared 195.67*** 161.25*** 162.88*** 198.91**
Likelihood-ratio test 0.1425 0.0775 0.2126 0.81
of rho = 0:

(3.1.1) included all the independent variables for dealers in the first equation, and then, all the independent variables for brands, and DLOY-
ALTY in the second
(3.2.1) included all the variables for dealers except EDUC, FSIZE, REL (see Chi-square test), and then, all the variables for brands except GS,
PRICE (see Chi-square test), and DLOYALTY
(3.3.1) included only the variables statistically significant for dealers in the models (2.1)–(2.2), and then, only the variables statistically signifi-
cant for brands in the models (1.1)–(1.2), and DLOYALTY
(3.4.1) included the variables for dealers in the model (2.3), and then, the variables for brands in the model (1.4), and DLOYALTY
Standard errors in (). (*), (**) and (***) indicate significance at the 0.10, 0.05 and 0.01, respectively
Bold letters are used to highlight the presence of one type of loyalty as an explanatory variable for another type of loyalty

serves as the dependent variable in the first stage, and as
an independent variable in the second stage, while Table 6
shows the models where brand loyalty acts as the dependent
variable in the first stage, and as an independent variable
in the following. For analysis purposes, the focus is on the
models (3.4.1) and (3.4.2) respectively because they are the
most parsimonious. At this stage of the analysis, it is crucial
to note that, according to the statistically significant coef-
ficients in both biprobit models, the same influential factors
on both types of loyalty detected in the univariate approach
appear again as the determinants in our bivariate approach.
That is, there is statistical consistency through our estima-
tions (see notes at the end of Tables 5 and 6, respectively).
The following is a crucial result of our research: produc-
ers considered loyal to agricultural machinery dealers will
have a 52.80% better chance of being brand loyal, com-
pared to producers regarded as disloyal to these dealers. On
the other hand, producers considered loyal to agricultural
machinery brands have a 45.82% better chance of being loyal
to dealers, compared to producers regarded as disloyal to the
brands. The two estimated parameters, DLOYALTY in the
model (3.4.1) and that of BLOYALTY in the model (3.4.2)
are statistically significant under any confidence level. Thus,
while dealer loyalty is approximately 7% above brand loy-
alty, in our modeling dynamics and under the context of
some kind of ‘empirical causality’, it is not possible to rank
the type of loyalty that influences the most the other loy-
alty. Therefore, it is possible to conclude that the statistical
influence of one type of loyalty over the other is strictly bi-
univocal. Reference can also be made to an ‘average loyalty
attraction’ in the agricultural machinery market under study
of approximately 49% on the part of producers.
Multivariate analysis for dual loyalty
To analyze the dual loyalty to brands and dealers alike, a
couple of multinomial probit models which results are
summarized in Table 7, were adjusted. However, our final
analysis is based on the second model (see note at the end
of Table 7).
It is a known fact that the factors determining Argen-
tinian producers’ loyalty to agricultural machinery brand
Producers’ brand‑dealer dual loyalty to capital equipment

Table 6  Seemingly unrelated bivariate probit models: first brand loyalty as the dependent, and then dealer loyalty as the dependent
Dependent variable Variable (3.1.2) (3.2.2) (3.3.2) (3.4.2) Marginal effects

BLOYALTY CONSTANT − 0.7262* − 0.6946* − 0.5153*** − 0.5397***

(0.3836) (0.3655) (0.1849) (0.1842)
EDUC − 0.2726** − 0.2743** − 0.2863** − 0.3198** − 0.1122
(0.1379) (0.1375) (0.1352) (0.1334)
AGEG2 − 0.5385*** − 0.5073*** − 0.5294*** − 0.5606*** − 0.1802
(0.1468) (0.1449) (0.1421) (0.1397)
AGEG3 − 0.8959*** − 0.9029*** − 0.9574*** − 0.9533*** − 0.2748
(0.1787) (0.1781) (0.1700) (0.1697)
RESID − 0.3114** − 0.2769** − 0.2170
(0.1437) (0.1414) (0.1330)
FSIZE2 0.0807 0.1811
(0.1585) (0.1425)
FSIZE3 0.1141 0.3121
(0.2384) (0.2003)
PRL 0.1072 0.0973
(0.1304) (0.1296)
GS2 0.0875
(0.1589)
GS3 0.2969
(0.1969)
CS 0.3092** 0.3110** 0.3293** 0.2972*** 0.1014
(0.1325) (0.1323) (0.1291) (0.1275)
DBEBP2 0.6386*** 0.6224*** 0.6476*** 0.6543*** 0.2451
(0.2028) (0.2007) (0.1966) (0.1962)
DBEBP3 0.3910*** 0.3793** 0.4361*** 0.4411*** 0.1531
(0.1526) (0.1524) (0.1433) (0.1432)
PRICE − 0.0742
(0.1890)
PERFORM 0.3922** 0.4282*** 0.4670*** 0.4652*** 0.1563
(0.1606) (0.1245) (0.1240) (0.1215)
MEDSOU 0.2198 0.1890
(0.1587) (0.1548)
MEANVAL − 0.0128 − 0.0150
(0.0886) (0.3655)
DLOYALTY CONSTANT − 0.3961 − 0.5096** − 0.4784* − 0.5207**
(0.3229) (0.2458) (0.2500) (0.2423)
BLOYALTY 1.1557** 1.2518*** 1.2168*** 1.2984*** 0.4582
(0.5126) (0.4382) (0.4244) (0.4193)
EDUC − 0.0670
(0.1398)
AGEG2 − 0.0494 − 0.0269 − 0.0388 − 0.0201
(0.1818) (0.1728) (0.1704) (0.1731)
AGEG3 − 0.0538 − 0.0022 − 0.0228 0.0049
(0.2234) (0.2056) (0.2073) (0.2060)
RESID − 0.1994 − 0.2010 − 0.2020 − 0.1954
(0.1368) (0.1326) (0.1341) (0.1298)
FSIZE2 − 0.0891
(0.1600)
FSIZE3 − 0.0781
(0.2442)
 P. Harmath et al.

Table 6  (continued)
Dependent variable Variable (3.1.2) (3.2.2) (3.3.2) (3.4.2) Marginal effects

PRL 0.3304** 0.3295** 0.3427*** 0.3392*** 0.1328

(0.1337) (0.1340) (0.1270) (0.1274)
GS2 0.4101*** 0.3890*** 0.3905*** 0.3941*** 0.1529
(0.1456) (0.1420) (0.1423) (0.1420)
GS3 0.2825 0.2108 0.2148 0.2172
(0.1955) (0.1584) (0.1577) (0.1578)
FPDR 0.0739 0.0704
(0.1451) (0.1448)
REL − 0.0594
(0.1468)
rho 0.1043 0.0494 0.0729 0.0212
(0.3006) (0.2625) (0.2517) (0.2533)
Log-likelihood − 612.3061 − 613.9158 − 616.9800 − 618.3201
Wald Chi-squared 145.26*** 146.35*** 137.10*** 137.63***
Likelihood-ratio test 0.1197 0.0352 0.0832 0.0070
of rho = 0

(3.1.2) included all the independent variables for brands in the first equation, and then, all the independent variables for dealers, and BLOY-
ALTY in the second
(3.2.2) included all the variables for brands except GS, PRICE (see Chi-square test), and then, all the variables for dealers except EDUC, FSIZE,
REL (see Chi-square test), and BLOYALTY
(3.3.2) included only the variables statistically significant for brands in the models (1.1)–(1.2), and then, only the variables statistically signifi-
cant for dealers in the models (2.1)–(2.2), and BLOYALTY
(3.4.2) included the variables for brands in the model (1.4.), and then, the variables for dealers in the model (2.3), and BLOYALTY
Standard errors in (). (*), (**) and (***) indicate significance at the 0.10, 0.05 and 0.01, respectively
Bold letters are used to highlight the presence of one type of loyalty as an explanatory variable for another type of loyalty

(BLOYALTY) are inherent characteristics of this market.
The following are to be outlined: producers’ age and educa-
tion level (EDUC, AGEG), their cognitive style (CS), their
appreciation towards product performance (PERFORM),
and their perception of any notable differences between
generic and branded products (DBEBP).
On the other hand, the factors that determine loyalty
to dealers (DLOYALTY) correspond to a combination
of the following: the producer’s age (AGEG), and three
other variables exposing a ‘symbiosis’ prototype between
the producer and his farm: the percentage of rented land
(PRL), his/her gross sales volume (GS), and the distance
between his/her farming operations and his/her residence
(RESID). The Eq. (4.2) for BDLOYALTY1 shows that
these are the factors that determine loyalty to brands and
dealers jointly or individually. There is an additional ele-
ment (factor): producers now use traditional media or
personal communication, which will probably have them
more likely to be considered loyal to both agricultural
machinery brands and dealers, compared to those produc-
ers who are more inclined to use media and social media
(MEDSOU) as a source of information.
It is interesting to note that the producer’s education
level and age are contrasted with his/her possible level of
loyalty, which is not the case with some aspects linked to
the producer’s beliefs, attitudes, and appreciations. Moreo-
ver, producers who live far from their farms may not have
a very close relationship with the agricultural machinery
dealer and may not be very familiar with the brands, which
may affect their loyalties. This is likely to narrow their
loyalty, as well as the possible links that may exist between
producers and various agricultural machinery brands and
dealers channeled through more traditional means of com-
munication than those currently used (social networks).
Finally, we cannot lose sight that, according to Eq. (4.2)
that the estimated BDLOYALTY2 and BDLOYALTY3, resi-
dence (RESID) is an important factor for producers who are
loyal to brands but not to dealers, and to those producers
who are otherwise loyal to dealers but not to brands. The
same is true for producers with midsize gross sales volumes
(GS2). In addition, the producer's cognitive style (CS) influ-
ences their loyalty to brands and not to dealers; while the
percentage of rented land (PRL) is an important factor for
producers’ loyalty to dealers, but not to brands.
Producers’ brand‑dealer dual loyalty to capital equipment

Table 7  Multinomial probit models for dual loyalty

Variable Dependent variable: BDLOYALTY
BDLOYALTY1 BDLOYALTY2 BDLOYALTY3
(4.1) (4.2) Marg. eff (4.1) (4.2) Marg. eff (4.1) (4.2) Marg. eff

CONSTANT − 0.6320 − 0.980*** − 1.6178* − 1.58*** − 0.1641 − 0.5841*

(0.5483) (0.3459) (0.7803) (0.4783) (0.5426) (0.3353)
EDUC − 0.3808* − 0.3599* − 0.0556 − 0.4433 − 0.1694
(0.2046) (0.2028) (0.2897) (0.2009)
AGEG2 − 0.674*** − 0.657*** − 0.1344 − 0.5210 − 0.0029
(0.2178) (0.2157) (0.3243) (0.2246)
AGEG3 − 1.159*** − 1.137*** − 0.2428 − 0.5983 0.0496
(0.2647) (0.2578) (0.3708) (0.2508)
RESID − 0.4838** − 0.4784** − 0.0623 − 0.4909 − 0.5655* − 0.0217 − 0.3191 − 0.3545* − 0.033
(0.2077) (0.2040) (0.3118) (0.3084) (0.1953) (0.1931)
FSIZE2 0.0243 − 0.1102 − 0.1776
(0.2371) (0.3375) (0.2380)
FSIZE3 0.0607 − 0.8959 − 0.4438
(0.3494) (0.6488) (0.3648)
PRL 0.3492* 0.3848** 0.0263 0.4522 0.6366*** 0.6130*** 0.1046
(0.1958) (0.1902) (0.2844) (0.2034) (0.1962)
GS2 0.3700 0.3837* 0.0145 0.6128* 0.5659* 0.0169 0.7537*** 0.7020*** 0.1216
(0.2355) (0.2240) (0.3192) (0.3054) (0.2183) (0.2104)
GS3 0.5765** 0.6084*** 0.1171 0.3851 0.3824
(0.2892) (0.2349) (0.4335) (0.2921)
CS 0.4090** 0.3937** 0.0551 0.7092** 0.6828** 0.0364 0.2021
(0.1964) (0.1940) (0.2939) (0.2872) (0.1996)
DBEBP2 0.9274*** 0.8976*** 0.1758 0.5996 0.1036
(0.3066) (0.3030) (0.4183) (0.3266)
DBEBP3 0.4918** 0.4479** 0.1129 0.0738 − 0.1730
(0.2269) (0.2207) (0.3288) (0.2323)
PERFORM 0.4650* 0.4886*** 0.1113 0.3768 − 0.1607
(0.1827) (0.1800) (0.2666) (0.1758)
MEDSOU 0.3954* 0.3885* 0.0922 − 0.1526 − 0.0056
(0.2255) (0.2239) (0.3098) (0.2096)
MEANVAL − 0.1109 0.0054 − 0.1342
(0.1280) (0.1869) (0.1297)
FPDR 0.1529 − 0.1174 − 0.0165
(0.2088) (0.3234) (0.2185)

BDLOYALTY4 is the base outcome. (4.1) included all the variables except PRICE, REL (see Chi-square test). Log- likelihood for
(4.1) = −  598.3216. Wald Chi-squared = 141.27***. (4.2) included only the variables statistically significant in (4.1). Log-likelihood for
(4.2) = − 601.7991. Wald Chi-squared = 135.95***
Standard errors in (). (*), (**) and (***) indicate significance at the 0.10, 0.05 and 0.01, respectively

Conclusions education, his cognitive style, his appreciation of the dif-

The following are the main conclusions of this work for the
Argentinian Ag equipment market. First, the results show
that even if brand loyalty and dealer loyalty to agricultural
capital equipment are related, the factors that determine
brand and dealer loyalty are not the same. Brand loyalty
depends on five explanatory variables: the farmer’s age and
ferences between expendable and branded products, and
product performance. Dealer loyalty, on the other hand, is
explained by the farmer’s age and residency, the percentage
of rented land, and his/her gross sales volume. No evidence
supporting the idea that the variables ‘relationship with
dealer’ and ‘price’ affect dealer loyalty and brand loyalty

was found. Also ‘farm size’ and ‘media sources’ do not seem
to explain either brand or dealer loyalty.
The results establish that farmers will tend to be more
loyal to capital equipment brands if they are young, have
received a less formal education, and are more analytic, if
they can perceive the difference between expendable and
branded products, and if they value the product performance.
Also, farmers will tend to be more loyal to the Ag capital
equipment dealer if they are younger, live near the farm, rent
more land, and have larger gross sales.
This first conclusion establishes that dealer loyalty is
more linked to demographic and farm characteristics, while
brand loyalty relates to a combination of demographic data
and farm, characteristics, farmers’ beliefs and attitudes,
product characteristics, and media exposure.
Second, considering how brand loyalty impacts dealer
loyalty and vice versa, we found that the influence of one
type of loyalty over the other is strictly bi-univocal: none of
the two loyalties predominate, as each needs the other. The
results show that among producers loyal to dealers they will
have a 52.80% better chance of being brand loyal, compared
to producers considered disloyal to these dealers. Producers
loyal to agricultural machinery brands will have a 45.82%
better chance of being loyal to dealers, compared to produc-
ers considered disloyal to these brands. However, it is not
possible to establish a hierarchy for which type of loyalty
exerts the greatest influence over the other.
This is consistent with comments from Ag equipment
managers we interviewed, who consider that in Argentina
the agricultural machinery purchasing process is a result of
a combination of the two loyalties. A good brand with a
poor dealership will negatively impact the brand, and a good
dealership without a brand has no sales.
This second conclusion determines that neither of the two
types of loyalty individually prevails over the other, and both
are required to attract customers and commit them to pur-
chase the Ag equipment. This result differs from the one
obtained by Harbor (2006), which concluded that the statisti-
cal factors that directly impact dealer loyalty may indirectly
impact brand loyalty.
Third, we also identified the factors that explain dual
loyalty, represented by the sum of factors explaining brand
loyalty and dealer loyalty plus the use of traditional media or
personal communication by farmers. This third conclusion
ratifies conclusions one and two, as it considers the same
factors that explain brand loyal and dealer loyalty separately,
adding another factor which is the personal and traditional
communication media.
This conclusion is conceptually similar to the one reached
by Harbor (2006) who established that activities that asso-
ciate the dealer with the brand are likely to help to build
dual loyalty (loyalty to brands and dealers) in the agricul-
tural machinery market. This would be analogous to what
P. Harmath et al.
happens in the car industry, which is an industrial market
that shares some similar characteristics with agricultural
machinery. However, unlike the automotive market, where
the customer goes to the dealer when he needs to make a
trade, in the agricultural machinery market the relationship
with the dealer is much closer, precisely because the farmer
is buying a capital good, which requires reviewing, repair-
ing, advising on use, etc.
This third conclusion establishes that achieving dual
loyalty is a very challenging goal, as the farmer fulfilling
this profile would have to sum up the factors explaining
both brand and dealer loyalty, when considered separately,
plus the preference for personal and traditional communi-
cation media.
Thus, the contribution of this paper is to have defined
the factors that explain brand and dealer loyalty for capi-
tal equipment, as well as the interplay between these two
loyalties, in the agribusiness context using probit models,
which has scarcely been researched up to now. This not
only contributes to the literature but also has interesting
business implications, such as that Ag equipment compa-
nies should invest in understanding the factors that drive
agricultural producers to purchase agricultural equipment.
This would help them to identify the purchasing factors
and segment farmers accordingly and develop appropriate
marketing strategies to attract different types of producers.
A second implication is the advantage for a given capital
equipment branded product to strengthen its distribution
channel, which calls for a cooperative relationship between
manufacturers and dealers. A third implication is that in
order to reach loyal farmers, Ag equipment companies
should identify the best communication channels. In the
Argentine context, agricultural producer mostly appreci-
ates the information received first-hand, face-to-face, and
the information obtained through the Internet and social
networks was the least appreciated by producers by the
time they were surveyed. However, this will surely change
in different contexts and time, which should be the object
of future studies.
The limitations of this work are related to the specific-
ity of the information used, as the data was obtained in a
2017 survey from a geographic region (Humid Pampa, in
Argentina), for agricultural producers who operate farms
of at least 200–300 hectares. Our estimations were made
in a specific time framework, they are of the cross-section
type, with information of several periods we will be able
to make panel estimations. Also, a large number of media
communication variables were included in this survey, and
time and technological changes tend to change farmers’
communication preferences.
Future research should be directed in search for panel-
based data to analyze this same situation for farmers who
were already surveyed, in order to see if the loyalty factors
Producers’ brand‑dealer dual loyalty to capital equipment

change over time. Likewise, the possibility of compar-

ing the behavior of this market with other similar ones in
other regions seems interesting, whether or not it has panel
data. Also, we could develop a conceptual framework to
analyze, from a qualitative perspective, how the differ-
ent dimensions of variables defined in Table 1 affect the
diverse types of loyalties in the agricultural machinery
market, similarly as done by Dhiman and Arora (2020) for

− 25.29***
− 10.81***

34.66***

10.13***

24.54***
2.09**
the hospitality industry in India. It would also be interest-
ing to analyze loyalty in the capital equipment market from

t-test
a value chain approach, following the work presented by
Ho et al. (2019).

Appendix A: Cluster analysis

for the construction of the dependent
variables

T-mean sample comparison test with unequal variances. (*), (**) and (***) indicate significance at the 0.10, 0.05 and 0.01, respectively
24% (100–76%)

79% (100–21%)

39% (100–61%)
% of response
As it has been done in previous work such as Roucan-Kane

Cluster 2
et al. (2011), we apply cluster analysis to identify groups of

n = 382
producers based on their loyalty to capital equipment brands

76%

26%

8%
and dealers, using the options available in questions 40 and
41 of the “The Needs of the Argentine Agricultural Pro-
ducer” questionnaire:

a. I will do more business with this brand/dealer.

Levene's test for equality of variances was performed for each variable and was significant at p < 0.01
b. I endorse this brand/dealer to my neighbors.
c. I try products from other brands/dealers.
d. I would switch to another brand/dealer to save 5%.
% of response

e. I would switch to another brand/dealer to save 10%.

Cluster 1
n = 182

f. I am loyal to this brand/dealer (I do not change brands/

100%

100%

100%

85%
68%

70%

dealers even if prices increase more than 10%).

Table 8  Cluster solution for capital equipment brand loyalty with two groups

To interpret the results, options c, d and e are considered

responses associated with disloyalty, since they show a dis-
position to switch brands (even when this switching may not
happen in practice). On the other hand, options a, b and f are
considered replies associated with loyalty, since they imply
that the producer is somehow involved with the brand (he
49% (100–51%)

86% (100–14%)

59% (100–41%)
% of response

would not switch even with higher prices, recommend it to

neighbors or seek to do more business with it).
n = 564
Total

We use a hierarchical clustering agglomeration method

33%
73%

40%

called linkage by averages between groups or inter-groups,

proposed by Sokal and Michener (1958). We calculate the
change brand if the price increases
I am loyal to this brand (I would not
DID NOT SELECT I would change

DID NOT SELECT I would change

DID NOT SELECT I try different

brand if the price increases 10%

similarities between the individuals in the sample based on

I will do more business with this

brand if the price increases 5%

I recommend this brand to other

the binary Euclidean distance since the variables selected

to carry out the cluster analyzes are all qualitative, each one
brands of this product

with two response categories, 0: no and 1: yes.

In Table 8, we observe a first group of 182 produc-
ers who will continue doing business with the brand and
Characteristic

show a strong commitment to it. They recommend the

farmers
brand

brand to their neighbors, and none of them try different

10%)

products nor will switch brands facing a 5% or even a



Table 9  Cluster solution for capital equipment dealer loyalty with two groups
Characteristic Total Cluster 1 Cluster 2 t-test
n = 564 n = 299 n = 265
% of response % of response % of response

I will do more business with this 74% 78% 69% − 2.38**

dealer
I recommend this dealer to other 47% 63% 28% − 8.87***
farmers
DID NOT SELECT I try different 57% (100–43%) 97% (100–3%) 13% (100–87%) 35.82***
dealers of this product
DID NOT SELECT I would change 88% (100–12%) 98% (100–2%) 77% (100–23%) 7.64***
dealer if the price increases 5%
DID NOT SELECT I would change 66% (1–34%) 83% (100–17%) 48% (100–52%) 9.29***
dealer if the price increases 10%
I am loyal to this dealer (I would not 27% 48% 3% − 14.43***
change dealer if the price increases
10%)

T-mean sample comparison test with unequal variances. (*), (**) and (***) indicate significance at the 0.10, 0.05 and 0.01, respectively
Levene's test for equality of variances was performed for each variable and was significant at p < 0.01
P. Harmath et al.
Producers’ brand‑dealer dual loyalty to capital equipment
10% price increase. Additionally, 85% of them claim that
they would stick to the brand even when facing a price
increase above 10%. This group of producers can be con-
sidered ‘empirically loyal’ to capital equipment brands.
On the other hand, a second group includes 382 produc-
ers, which are ‘empirically disloyal’ to capital equipment
brands. With respect to dealer loyalty, as we see in Table
9, the first cluster is a group of 299 producers, whom we
call ‘empirically loyal’ to the dealer. In the second cluster,
there is a different group of producers, whom we call the
‘empirically disloyal’.
In both cluster solutions, we may see that the t-tests for
percentage comparison (assuming that the population vari-
ances are different) indicate that the differences for each
variable between the groups are statistically significant.
Therefore, we can conclude that there are two significantly
different groups in terms of brand loyalty and dealer loy-
alty, as defined by the combination of six proposed vari-
ables; that is, the cluster solutions are statistically valid.
Consequently, in terms of dual loyalty, we identify 156
producers considered loyal to both brands and dealers, 26
loyal to brands and disloyal to dealers, 143 loyal to deal-
ers and disloyal to brands, and 239 producers considered
disloyal to both brands and dealers.
Acknowledgements  Not applicable.
Author contributions  All authors contributed to the design of the work.
PH and JRP analyzed and interpreted data. RF and PH drafted the work.
JRP revised it. All authors read and approved the final manuscript.
Funding  Not applicable.
Availability of data and materials  The dataset supporting the conclu-
sions of this article is available from the Universidad Austral. Restric-
tions apply to the availability of this data, which were used under
license for this study. Data are available from the authors with the
permission of the Universidad Austral.
Declarations  ers’ management practices and strategies: A comparison
Competing interests  The authors declare that they have no competing
interests.
References
Ag Equipment Intelligence. 2018. Brand Loyalty in the Farm Equip-
ment Business. https://​www.​farm-​equip​ment.​com/​ext/​resou​rces/​
aei/​Repor​ts/​AEI_​Brand-​Loyal​ty-​in-​the-​Farm-​Equip​ment-​Busin​
ess_​Report_​0418_​web.​pdf. Accessed 6 Mar 2021.
Akaeze, C.O., and N.A.S. Akaeze. 2017. Exploring factors that influ-
ence consumer loyalty to automotive dealership in New York.
Journal of Business Theory and Practice 5: 98–119. https://​doi.​
org/​10.​22158/​jbtp.​v5n2p​98.
Anisimova, T., and F. Mavondo. 2014. Aligning company and dealer
perspectives in corporate branding: Implications for dealer satis-
faction and commitment. Journal of Business-to-Business Market-
ing 21: 35–56. https://​doi.​org/​10.​1080/​10517​12X.​2014.​857501.
Cameron, A.C., and P.K. Trivedi. 2005. Microeconometrics: Methods
and applications. Cambridge: Cambridge University Press.
Cassia, F., N. Cobelli, and M. Ugolini. 2017. The effects of goods-
related and service related B2B brand images on customer loyalty.
Journal of Business & Industrial Marketing 32: 722–732. https://​
doi.​org/​10.​1108/​JBIM-​05-​2016-​0095.
Dhiman, P., and S. Arora. 2020. A conceptual framework for identify-
ing key employee branding dimensions: A study of hospitality
industry. Journal of Innovation & Knowledge 5 (3): 200–209.
https://​doi.​org/​10.​1016/j.​jik.​2019.​08.​003.
Ewing, M.T. 2000. Brand and retailer loyalty: Past behavior and future
intentions. The Journal of Product and Brand Management 9:
120–127.
Greene, W.H. 2012. Econometric analysis. Hoboken: Prentice-Hall
Publishing, Pearson Education Limited.
Glynn, M.S. 2010. The moderating effect of brand strength in manufac-
turer–reseller relationships. Industrial Marketing Management 39:
1226–1233. https://​doi.​org/​10.​1016/j.​indma​rman.​2010.​02.​019.
Harbor, A. 2006. Assessing agricultural input brand loyalty among
United States mid-size and commercial producers. Doctoral dis-
sertation, Purdue University, West Lafayette, Indiana.
Ho, K.L.P., C.N. Nguyen, R. Adhikari, M.P. Miles, and L. Bonney.
2019. Leveraging innovation knowledge management to cre-
ate positional advantage in agricultural value chains. Journal
of Innovation & Knowledge 4 (2): 115–123. https://​doi.​org/​10.​
1016/j.​jik.​2017.​08.​001.
Huber, F., and A. Herrmann. 2001. Achieving brand and dealer loy-
alty: The case of the automotive industry. The International
Review of Retail, Distribution and Consumer Research 11:
97–122. https://​doi.​org/​10.​1080/​71377​0584.
Kool, M., M. Meulenberg, and D. Browns. 1997. Extensiveness of
farmers’ buying processes. Agribusiness 13: 301–318.
Lynch, J., and L. De Chernatony. 2004. The power of emotion-brand
communication in business-to-business markets. Journal of
Brand Management 11: 403–419. https://​doi.​org/​10.​1057/​palgr​
ave.​bm.​25401​85.
Leek, S., and G. Christodoulides. 2012. A framework of brand
value in B2B markets: The contributing role of functional and
emotional components. Industrial Marketing Management 41:
106–114. https://​doi.​org/​10.​1016/j.​indma​r man.​2011.​11.​009.
Mohammadi, Z.M., P. Mac Clay, R. Feeney, P. Harmath, M. Kes-
havarz, and M.A. Gunderson. 2020. Characterization of farm-
between Argentine and U.S. farmers. International Food and
Agribusiness Management Review 23: 235–251.
Mudambi, S. 2002. Branding importance in business-to-business
markets: Three buyer clusters. Industrial Marketing Manage-
ment 31: 525–533. https://​d oi.​o rg/​1 0.​1 016/​S 0019-​8 501(02)​
00184-0.
Nyström, A.G., J.A. Törnroos, N. Koporcic, and M. Ivanova-Gongne.
2018. Summary: Branding in the B2B context and future chal-
lenges. In Developing insights on branding in the B2B context,
ed. N. Koporcic, M. Ivanova-Gongne, A.G. Nyström, and J.A.
Törnroos, 219–229. Bingley: Emerald Publishing Limited. https://​
doi.​org/​10.​1108/​978-1-​78756-​275-​22018​1013.
Reportlinker. 2019. The global agriculture equipment market size is
expected to reach USD 244.2 billion by 2025, expanding at a
CAGR of 8.9% from 2019 to 2025. https://w ​ ww.p​ rnews​ wire.c​ om/​
news-​relea​ses/​the-​global-​agric​ulture-​equip​ment-​market-​size-​is-​
expec​ted-​to-​reach-​usd-​244-2-​billi​on-​by-​2025--​expan​ding-​at-a-​
cagr-​of-8-​9-​from-​2019-​to-​2025--​30079​4964.​html. Accessed 6
Mar 2021.

Roucan-Kane, M., C. Alexander, M. Boehlje, S. Downey, and A. Gray.
2011. A large commercial producer market segments for agri-
cultural capital equipment. International Food and Agribusiness
Management Review 14: 127–143.
Seyedghorban, Z., M.J. Matanda, and P. La Placa. 2016. Advancing
theory and knowledge in the business-to-business branding lit-
erature. Journal of Business Research 69: 2664–2677. https://​doi.​
org/​10.​1016/j.​jbusr​es.​2015.​11.​002.
Seyoum, S. 2018. Analysis of prevalence of malaria and anemia using
bivariate probit model. Annals of Data Science 5: 301–312.
https://​doi.​org/​10.​1007/​s40745-​018-​0138-3.
Sokal, R.R., and C.D. Michener. 1958. A statistical method for evaluat-
ing systematic relationship. University of Kansas Scientific Bul-
letin 28: 1409–1438.
Verhoef, P.C., F. Langerak, and B. Donkers. 2007. Understanding brand
and dealer retention in the new car market: The moderating role
of brand tier. Journal of Retailing 83: 97–114. https://​doi.​org/​10.​
1016/j.​jretai.​2006.​10.​007.
Viardot, E. 2017. Branding in B2B: The value of consumer goods
brands in industrial markets. Journal of Business and Indus-
trial Marketing 32: 337–346. https:// ​ d oi. ​ o rg/ ​ 1 0. ​ 1 108/​
JBIM-​11-​2014-​0225.
Walley, K., P. Custance, S. Taylor, A. Lindgreen, and M. Hingley. 2007.
The importance of brand in the industrial purchase decision: A
case study of the UK tractor market. Journal of Business & Indus-
trial Marketing 2: 383–393. https://​doi.​org/​10.​1108/​08858​62071​
07801​45.
Wooldridge, J.M. 2010. Econometric analysis of cross section and
panel data. Cambridge: The MIT Press.
Hinterhuber, A., and Hinterhuber, G. 2012. An empirical analysis of
the role of industrial brands for industrial distributors. Journal of
P. Harmath et al.
Strategy & Management 5(3): 252–265. https://​doi.​org/​10.​1108/​
17554​25121​12475​62
Publisher's Note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.
Pedro Harmath  Associate Professor and researcher of the School of
Business Sciences, Department of Mathematics, Universidad Austral,
Argentina. He has a Ph.D. in Mathematics (Universidad Centrocciden-
tal Lisandro Alvarado), a Master’s in Quantitative Economics (Univer-
sidad de Los Andes) and a Bachelor’s Degree in Statistics (Universidad
de Los Andes). He does research in applied statistics in the fields of
agricultural economics and behavioral economics.
Roberto Feeney  Associate Professor of the Center for Food and Agri-
business, Universidad Austral, Argentina. His educational background
includes a Ph.D. in Business Administration, MBA, and a bachelor’s
degree in Economics. His main research interest is on topics such as
innovation management, bio-based economics, Ag value chains, and
behavioral economics and agribusiness decision making.
Josefa Ramoni‑Perazzi  Full Professor of the Escuela de Economía,
Universidad Industrial de Santander (Bucaramanga, Santander, Colom-
bia). She has a Bachelor´s Degree in Economics (Venezuela), M.Sc.
in Statistics (Venezuela) and in Economics (USA), and a PhD. in Eco-
nomics (USA). Member of the research group in Regional Develop-
ment and Territorial Planning (GIDROT-UIS) and Study Group on
Applied Microeconomics and Regulation (EMAR-UIS).