Content uploaded by Jose Maria Martínez
Author content
All content in this area was uploaded by Jose Maria Martínez on May 14, 2021
Content may be subject to copyright.
RESEARCH ARTICLE
Joint adoption of rice technologies among
Bolivian farmers
Jose Maria Martinez1, Ricardo A. Labarta2, Carolina Gonzalez2and Diana C. Lopera2,3
1
Department of Agricultural, Food, and Resource Economics, Michigan State University, East Lansing, MI,
USA,
2
Alliance of Bioversity International and CIAT, Cali, Valle del Cauca, Colombia and
3
HarvestPlus,
Cali, Valle del Cauca, Colombia
Corresponding author: J. M. Martinez. Email: mart2388@msu.edu
(Received 31 July 2020; revised 19 April 2021; accepted 19 April 2021)
Abstract
Bolivia has disseminated several improved technologies in the rice sector, but the average
rice productivity in the country is far below the average trend in Latin America in recent
years. Although the economic literature has highlighted the role of agricultural technology
adoption in increasing agricultural productivity, gaps remain in understanding how rice
growers are deciding to adopt and benefit from available improved rice technologies.
Most previous adoption studies have evaluated the uptake of individual technologies with-
out paying attention to the complementarities that alternative improved rice technologies
may offer to farmers who face multiple marketing and production needs. This study uses
data from a nationally representative sample of Bolivian rice growers to analyze farmers’
joint decisions in adopting complementary agricultural technologies controlling for
potential correlations across these decisions, as well as the extent of adoption of these
practices. Evidence suggests that the decisions on multiple technology adoption are closely
related, with common factors affecting both adoption and the extent of adoption.
Furthermore, there is a need to better target resource-poor farmers, improve
information-diffusion channels on agricultural practices, and better use existing farmers’
organizations to enhance rice technology adoption.
Keywords: Bolivia; multivariate Probit; rice; technology adoption
JEL: O13; Q12; Q16
Introduction
Rice productivity in Latin America has shown a considerable increase over recent
decades. However, these productivity gains have not been homogeneous across the rice-
producing countries in the region. As of 2018, countries such as Argentina, Paraguay,
Peru, and Uruguay, and Brazil’s southern region, had achieved rice yields between 6.3
and 8.5 t/ha, whereas countries such as Bolivia and Panama remained at production
rates below 4 t/ha. In Bolivia, rice production is an essential source of farmers’income
© The Author(s), 2021. Published by Cambridge University Press on behalf of the Northeastern Agricultural and Resource
Economics Association. This is an Open Access article, distributed under the terms of the Creative Commons Attribution
licence (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution, and reproduction in
any medium, provided the original work is properly cited.
Agricultural and Resource Economics Review (2021), 1–21
doi:10.1017/age.2021.9
https://doi.org/10.1017/age.2021.9
Downloaded from https://www.cambridge.org/core. IP address: 72.44.113.147, on 14 May 2021 at 15:02:45, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
and a critical factor in achieving food security (Calpe 2006). Yet, this country has had
the lowest historical yield and second-lowest productivity growth in the region
(FAOSTAT 2020).
The economic literature has highlighted the role of agricultural technology adoption
in increasing farm productivity and has identified various factors that explain the adop-
tion of different technologies (Feder, Just, and Zilberman 1985; Foster and Rosenzweig
2010). Most studies so far have concentrated on explaining successful technology adop-
tion processes but less on explaining why technology adoption has not successfully
reached a larger target population and, therefore, has not translated into considerable
yield increases. While many technologies have been promoted in the rice sector of
Bolivia and other countries with low rice productivity, few studies have analyzed the
factors that may explain the unsuccessful adoption of agricultural technologies.
Studies on the adoption of agricultural technologies have somewhat focused on
improved crop varieties. Furthermore, despite the recognized complementarity of
improved varieties and other enhanced agronomic practices, most studies have mainly
examined separately the decision to adopt agronomic practices from the selection of
improved varieties. However, a growing body of the agricultural economics literature
has recognized the importance of modeling jointly different decisions in the agricultural
technology adoption process to account for potential correlation in alternative
adoption decisions (Wu and Babcock 1998). Teklewold, Kassie, and Shiferaw (2013)
modeled the adoption of sustainable agricultural practices among Ethiopian maize
farmers as a joint process while also incorporating the extent of such adoption into
the analysis—measured as the number of practices implemented by farm households
(following Wollni, Lee, and Thies (2010)). The authors show that the complementari-
ness and substitutability of agricultural practices are worth analyzing to better under-
stand the drivers of adoption and overall uptake. Building on Teklewold, Kassie, and
Shiferaw (2013), other studies have further provided evidence on the suitability of
model joint decisions about adopting agronomic practices and improved crop varieties
in China (Zeng et al. 2020), Ethiopia (Yirga, Atnafe, and AwHassan 2015; Gebremariam
and Tesfaye 2018), Ghana (Donkoh, Azumah, and Awuni 2019), India (Aryal et al.
2018), Kenya (Wainaina, Tongruksawattanab, and Qaima 2016; Kanyenji et al. 2020),
Malawi (Ward et al. 2018), Nigeria (Oladimeji et al. 2020), Tanzania (Kassie et al.
2013), and the Chinyanja Triangle (Mponella, Kassie, and Tamene 2018).
This article studies how rice growers in Bolivia make decisions about adopting dif-
ferent agricultural technologies, namely, modern improved varieties (MIVs), mechani-
zation, inorganic fertilization, pesticides, and herbicides. We allow these decisions to be
pair-wise correlated and test for the existence of any underlying complementarity.
Moreover, the extent of adoption is also modeled as an ordered response, further ana-
lyzing how technology adoption factors affect their overall uptake. Using a nationally
representative sample of rice growers collected during 2014, the article (a) tests the suit-
ability of using a multivariate model approach for studying the factors that affect the
adoption of rice technologies in Bolivia, (b) models the drivers on the extent of adop-
tion of these technologies, and (c) derives policy implications and recommendations to
better disseminate the rice technologies that would enhance the rice sector. Our results
build upon previous findings by providing valuable insights for the Latin America and
the Caribbean region. These kinds of analyses are virtually absent in LAC, as most stud-
ies with similar approaches highly concentrate on African cases.
Our research findings provide evidence of a significant joint correlation between the
different agricultural technologies disseminated in Bolivia’s rice sector. Likewise,
2 Martinez et al.
https://doi.org/10.1017/age.2021.9
Downloaded from https://www.cambridge.org/core. IP address: 72.44.113.147, on 14 May 2021 at 15:02:45, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
evidence suggests that distance to technological diffusion centers, participation in pro-
ducers’organizations, and access to extension services significantly affect the probability
of adopting these technologies. We also argue that, to better target resource-poor farm-
ers, there is a need to improve mechanisms to diffuse information on agricultural prac-
tices and promote better use of existing farmers’organizations to boost rice technology
adoption and improve the overall conditions of rice-producing households in the
country.
The article proceeds as follows: section two provides an overview of the Bolivian rice
sector, while section three describes the data and empirical strategy used in the analysis.
The final two sections, respectively, discuss the main results and summarize the conclu-
sions and policy recommendations.
Background: an overview of the Bolivian rice sector
Rice is the third most consumed food in Bolivia (GRiSP 2013). It constitutes a primary
source of dietary nutrients for its population and has become one of the major income-
generating crops in its agricultural sector. However, the country has had considerably
low historical yields compared with neighboring countries (Table 1), which has had
implications for both consumers’and rice farming households’welfare. In countries
Table 1. Rice ( paddy) yield metrics for Latin America and the Caribbean, 1960–2018
Yield annual growth rate
a
(%) Overall yield
b
Country 1961–1980 1981–2000 2001–2018 1961–2018 1961–2018 2014 2018
Argentina −0.29 2.15 0.92 1.43 5.23 6.50 6.90
Bolivia 0.11 1.52 1.58 1.19 2.14 2.77 3.19
Brazil −0.67 3.76 4.12 2.67 2.31 5.20 6.31
Colombia 5.36 0.05 0.38 1.25 4.24 4.78 5.22
Costa Rica 4.29 2.87 1.06 1.96 3.08 4.88 4.55
Dominican Republic 3.26 0.68 −2.30 1.23 3.99 3.26 3.26
Ecuador 2.26 1.23 0.10 1.26 3.53 3.90 4.53
Guyana 1.48 1.47 −0.02 1.74 3.33 3.43 5.77
Nicaragua 3.07 −0.38 4.40 1.37 3.71 5.58 5.27
Panama 2.76 1.65 0.73 2.09 2.14 3.26 3.34
Paraguay −1.54 4.55 4.60 2.03 4.37 6.70 6.30
Peru 0.39 1.89 1.10 1.39 6.14 7.60 8.12
Suriname 1.93 −0.33 1.57 0.57 3.88 4.43 4.87
Uruguay 1.67 1.26 1.87 1.81 6.27 8.05 8.50
Venezuela 4.21 3.85 −1.02 2.08 3.79 5.11 4.16
a
The annual growth rates are calculated from the linear regression of log-yields over the years, i.e., this is 100*β, where β
is coefficient from the linear regression log-( yield) = α+β*(year).
b
This is the simple average of yields for each country for the period 1960–2018, or for the specific years 2014 and 2018.
Source: Elaborated by the authors based on FAOSTAT (2020).
Agricultural and Resource Economics Review 3
https://doi.org/10.1017/age.2021.9
Downloaded from https://www.cambridge.org/core. IP address: 72.44.113.147, on 14 May 2021 at 15:02:45, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
such as Bolivia, Peru, and Ecuador, the rice-producing area is concentrated among
farms with less than 10 hectares (ha). By 2007, Bolivia had 43,456 rice-growing house-
holds, for a total area of 179,162 ha (Ortiz and Soliz 2007). Only a small proportion of
rice growers (1,157) had irrigation systems on a total area of 18,000 ha. Most Bolivian
rice production is under rainfed conditions. As expected, average yields differ signifi-
cantly between rainfed and irrigated systems. On average, farmers in rainfed rice pro-
duction reach 2 t/ha, while a limited number of farmers using irrigation in their rice
production reach more than 6 t/ha (Ortiz and Soliz 2007).
Bolivian rice growers, and the broader agricultural sector, have faced adverse local
market fluctuations and production challenges associated with a changing climate dur-
ing the past few decades (Amemiya 2001; Andersen and Verner 2009; Winters 2012).
Under this situation, and considering the role of rainfed production in the rice sector,
access to appropriate crop technologies such as improved rice varieties has become
more relevant, especially for small- and medium-scale farmers. Therefore, we focus
our analysis on these two groups of rice growers. If engaged in adopting new tech-
nologies, they may have better chances to increase their rice productivity and,
through higher yields and output quality, potentially have higher income and better
livelihoods.
Farm-gate real prices for rice have steadily decreased during the last three decades in
Bolivia and the main rice-producing countries in Latin America. Although this trend
translates into cheaper food in urban areas, the decline in rice growers’prices has
been higher in Bolivia. This has implied harsher conditions for small- and medium-
scale farmers, namely those who have failed to decrease their production costs to a sig-
nificant scale (i.e., low yields and declining prices have affected rice growers’expected
returns). Therefore, there are incentives for these groups of farmers to adopt new tech-
nologies for improving food security and productivity in Bolivia (Salazar et al. 2015).
However, at the same time, low crop profitability may discourage these farmers from
investing in these technologies. This article seeks to answer this empirical question.
Research on agricultural technologies in Bolivia is limited due to the low priority of
the farming sector and Bolivian governments’preference to support tin mining. Thus,
agriculture has been encouraged only in valleys because of the natural constraints faced
in other areas such as the highlands and tropical regions (Godoy, Morduch, and Bravo
1998). In recent decades, however, interest has increased in agriculture and in further
developing and promoting the use of improved technologies (Ortiz and Soliz 2007).
Rice is not a native crop to Latin America; farmers have only dealt with improved cul-
tivars, ruling out the use of landraces or local varieties. However, Bolivian farmers have
access to old and MIVs, introduced or officially released in the country at different
times (Ortiz and Soliz 2007; Labarta et al. 2014). The dissemination of other rice tech-
nologies has also been spread out over different periods.
Latin America had not been part of the Green Revolution until the 1980s, when
more international agricultural research centers were established worldwide
(Dalrymple 1986). By 1983, Brazil, Colombia, Peru, Venezuela, and Cuba had become
the region’s largest rice-growing countries. At that time, little was known about high-
yielding rice varieties (HYVs) in Bolivia. Still, a rough estimated area on HYVs reported
approximately 29,600 ha, mainly in upland and rainfed production systems. Besides
this, the use of other agricultural inputs in rice production in the 1970s was reported
to be limited (Dalrymple 1979).
By 1990, genetic resources for rice improvement in Bolivia had been virtually non-
existent (Taboada et al. 2005), and the production of this cereal had relied primarily on
4 Martinez et al.
https://doi.org/10.1017/age.2021.9
Downloaded from https://www.cambridge.org/core. IP address: 72.44.113.147, on 14 May 2021 at 15:02:45, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
old varieties such as “Bluebonnet 50,”introduced from the United States (Nguyen and
Tran 2002). In 1997, the Centro de Investigación Agrícola Tropical, in the Santa Cruz
region, started a rice genetic improvement program aimed at developing varieties better
adapted to local soil conditions and rainfed production systems (Taboada et al. 2005).
After ten years of work, they released the first two varieties developed for Bolivia. On
the one hand, biofortified variety “Azucena”was made available to local farmers
(HarvestPlus 2009) and, on the other hand, a collaboration with the Latin American
Fund for Irrigated Rice (FLAR) developed variety “MAC 18.”Six years later, two
other varieties (SACIA FL-39 and SACIA FL-40) were released, thus increasing the
number of modern improved rice varieties in the system, suitable for either manual,
mechanized, or irrigated systems.
Finally, a key element to further understand Bolivia’s rice production and technology
adoption is Japanese-descendant farmers’historical role. Since the first wave of Japanese
immigration into Bolivia during World War II, their influence over the Bolivian agri-
cultural sector has been significant. Japanese-descendant farmers have led both in
crop specialization and in the introduction of modern technologies in rice, soybean,
and wheat in their settling-in areas. Two of the initial Japanese settlements (Okinawa
and San Juan de Yapacaní) have become major technology dissemination points and
crop commercialization centers in the Bolivian rice sector (Amemiya 2001). Thus,
rice producers in the country could have benefited from being exposed to the
Japanese-descendant centers that implemented and developed best crop production
practices.
Materials and methods
Data
The data used for the analysis are a nationally representative sample of rice growers col-
lected during 2014 by the International Center for Tropical Agriculture (CIAT) in col-
laboration with the Centro de Investigación Agrícola Tropical of Santa Cruz, Bolivia,
across the three rice-producing regions in the country. The sampling design followed
a multistage procedure distributed among 98 communities, consisting of 802 small-
and medium-scale farmer households that provided the complete information required
under the proposed model.
Dependent variables
As mentioned earlier, rice is not a native crop of Latin America, and all varieties grown
by rice growers in Bolivia are genetically improved. However, the modern varieties
released since 2004 currently occupy 45.6 percent of the total rice area. The other
54.4 percent of Bolivia’s rice area uses old varieties mainly introduced from other coun-
tries before the foundation of the national rice breeding program. The rate of adoption
of the different MIVs is shown in Table 2. Two MIVs dominate the preference of
Bolivian rice growers: MAC 18, which resulted from the collaboration of the Bolivian
rice program and FLAR, and IAC 101, which was bred by the rice breeding program
of the State of São Paulo (Brazil), selected, and then released in Bolivia. These two vari-
eties have adapted well to both irrigated and favored rainfed conditions in Bolivia. MAC
18 also offers a slightly higher quality of its grain, which is appreciated in the market.
Despite being available to farmers for some time and expected to significantly improve
Agricultural and Resource Economics Review 5
https://doi.org/10.1017/age.2021.9
Downloaded from https://www.cambridge.org/core. IP address: 72.44.113.147, on 14 May 2021 at 15:02:45, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
productivity either in input-intensive (and mechanized) or manual production systems
(Châtel et al. 2010), the rest of the modern varieties have not achieved a considerable
rate of adoption. This is consistent with literature findings that productivity gains
may not necessarily cover the additional costs from adopting new crop varieties,
hence limiting their appeal to most farmers (Pretty, Ruben, and Thrupp 2002).
Besides, materials that have long been used in local conditions are likely to be better
adapted to their specific agroclimatic needs than their more recent competitors
(Hoffman, Probst, and Christinck 2007).
Although rainfed systems dominate rice production in Bolivia, large machinery for
planting and harvesting is available to rice growers. The possibility of mechanization
among rainfed rice growers refers to the concept of favored rainfed systems (Lynch
and Tasch 1981). Fertilization in rice production mainly relies on the use of mineral
supplements with a range of combinations of nitrogen (N), phosphorus (P), and potas-
sium (K). The most available fertilizer formulations in local markets offer N, P, and K
concentrations ranging from 21 to 46 percent, 20 to 45 percent, and 50 to 60 percent,
respectively (Viruez and Taboada 2013). Chemical pest control is virtually a binary
decision between products containing either cypermethrin or cyfluthrin to combat
both the fall armyworm and Tibraca limbativentris. In contrast, most weed controls
(herbicides) are made of bispyribac-sodium, imazapyr, imazapic, and penoxsulam.
We expect that rice growers in Bolivia have up to five technologies available that may
increase their rice productivity compared with traditional rice practices, but this may
also imply some degree of complementarity. These technologies, the adoption of
MIVs, machinery used at seeding or harvesting, and the use of fertilizers, chemical pes-
ticides, and herbicides on rice plots, constitute the dependent variables of our model.
Our definition of “modern”improved variety considers varieties released in Bolivia
since 2004 (eight years after establishing the rice improvement program). These are
Table 2. Adoption
a
of modern improved rice varieties in Bolivia
Variety Overall adoption (%) Adoption within adopters (%) Year of release
Amboró 0.12 0.25 2004
Azucena 0.12 0.25 2009
CAISY-50 0.24 0.51 2008
CICA-8 0.12 0.25 2009
Cristal 3.79 8.08 2008
Esperanza 0.24 0.51 2006
IAC 101 14.08 30.05 2005
MAC 18 25.09 53.54 2008
Paitití 4.38 9.34 2004
Saavedra 27 0.36 0.76 2009
Saavedra 44 0.59 1.26 2009
Tapeque 0.59 1.26 2004
a
Farmers may adopt more than one rice variety; hence, the reported values of adoption do not add up to 100 percent.
Source: Elaborated by the authors based on survey data.
6 Martinez et al.
https://doi.org/10.1017/age.2021.9
Downloaded from https://www.cambridge.org/core. IP address: 72.44.113.147, on 14 May 2021 at 15:02:45, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
up to 12 modern materials, including MAC 18, IAC 101, Azucena, SACIA FL-39, and
SACIA FL-40. Other varieties, including Bluebonnet, are considered old types.
Although these technologies could be expected as (nearly) perfect complements
from an agronomic perspective, the observed extent of adoption (Figure 1) suggests
that farms make only a partial adoption of such practices. Such a pattern is under-
standable for resource-limited households, whose decisions on production and
consumption are highly interdependent. The latter follows the concept of nonse-
parability of decisions of agricultural households (Bardhan and Udry 1999). It is
reasonable for these farms to adopt a part of an improvement package to gain at
least a part of the potential benefits from adoption. Households may possess attri-
butes that should be explored to understand better who are more likely to adopt
each technology of interest. Also, it is worth noting that although these technolo-
gies may be labeled as input-intensive, they are not necessarily incompatible with
future efforts in expanding the uptake of sustainable practices (Wainaina,
Tongruksawattanab, and Qaima 2016).
Econometric approach
In this study, we consider a farmer household deciding whether or not to adopt a given
technology. With no loss of generality, suppose that if household ichooses to adopt the
mth technology, it reaches a utility of V1
im, while it would reach a utility of V0
im
otherwise. Therefore, they adopt the technology if and only if
A∗
im=V1
im −V0
im =x′
ibm+
1
im .0, (1)
Figure 1. Number of Technologies Adopted by Interviewed Bolivian Rice Farmers. Source: Elaborated by the
authors based on survey data.
Agricultural and Resource Economics Review 7
https://doi.org/10.1017/age.2021.9
Downloaded from https://www.cambridge.org/core. IP address: 72.44.113.147, on 14 May 2021 at 15:02:45, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
where the ε
im
terms are unobservable shocks to the farmer utility for each mtechnol-
ogy, and x
i
are observed attributes of the household. Households are to choose whether
to adopt each of the m=1,2,…,Mnonexclusive technologies. While we do not observe
A∗
im or Vh
im, we do observe the adoption decision of each household as an index func-
tion, where Aim =1(A∗
im .0). Relaxing the assumption that the unobservable shocks
are independent, and assuming that they are jointly and normally distributed, leads
to the multivariate Probit (MVP) regression model (Cameron and Trivedi 2005;
Greene 2012), where
Aim =Pr (A∗
im .0) =Pr (−
1
im ,x′ibm)=Fm
M(x′ibm).(2)
Let us model the joint probability of adoption for Mtechnologies as
Pr (A1=1, A2=1, ...,AM=1) =x′b1
−1x′b2
−1
...x′bM
−1
fM(t,S)dt,(3)
where ϕ
M
is the standardized multivariate normal probability density function, and Σis
the symmetric covariance matrix of the error vector ε, in which Σ
mn
=1 if m=n, and
Σ
mn
=ρ
mn
=ρ
nm
when m≠n. Estimating the M-dimensional normal integrals is feasible
by the Geweke, Hajivassiliou, and Keane (GHK) simulated-likelihood approach
(Boersch-Supan and Hajivassiliou 1990; Cameron and Trivedi 2005; Greene 2012).
The decision to model adoption at the household level
1
follows that 95 percent of
the interviewed households managed a single plot. Barely over 1 percent managed
three or more plots. Thus, a household-level analysis does not conflict with similar
studies as those of Donkoh, Azumah, and Awuni (2019), Gebremariam and Tesfaye
(2018), and Zeng et al. (2020).
In addition to this, we model the extent of adoption as the number of technologies
effectively implemented by the household (Wollni, Lee and Thies 2010), setting up an
ordered Probit regression model that also resembles a random utility model. Here, farm
household idecides to adopt a total of c
i
technologies (c
i
= 0, 1, 2, 3, 4, 5) based on the
utility
Ui=h(x′iu)+
h
i.(4)
Although utility cannot be observed, as in the MVP setting, we observe the number of
adopted technologies. We assume that a household adopts an additional technology if the
utility of using it exceeds the utility of not adopting it. Moreover, the decision of extent of
adopted technologies responds to utility levels defined across thresholds λsuch that
ci=0, if Ui≤l1,(5)
ci=k,iflk,Ui≤lk+1for k=1, 2, 3, 4,
ci=5, if Ui.l5.
1
Additional results showed that the results hold when analyzing data at a plot level (Supplementary
Tables S1–S3), but we decided to keep the analysis at a household level as this is the unit of interest for
derived policy recommendations.
8 Martinez et al.
https://doi.org/10.1017/age.2021.9
Downloaded from https://www.cambridge.org/core. IP address: 72.44.113.147, on 14 May 2021 at 15:02:45, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
Finally, if ηfollows a normal distribution, then
Pr (ci=0|xi)=F(l1−x′iu), (6)
Pr (ci=k|xi)=F(lk+1−x′i
u
)−F(lk−x′iu), if k=1, 2, 3, 4,
Pr (ci=5|xi)=1−F(l5−x′iu).
The sign of θcoefficients in the ordered Probit model indicates the aggregate direc-
tion that covariates have in the number of technologies adopted by the household, i.e.,
whether the explanatory variable increases (decreases) the value of our latent variable U
(Cameron and Trivedi 2005). Nonetheless, marginal effects do not necessarily hold the
same sign, as covariates may imply more (or less) likelihood of a point outcome.
Explanatory variables and descriptive statistics
Covariates used in this article (Table 3) are consistent with previous adoption studies on
rice technologies and follow the underlying theoretical economic models that propose
them (Saka et al. 2005; Usman, Ango, and Barau 2013; Yamano et al. 2016). We have
included producers’associations and access to extension services to represent house-
holds’access to agricultural information. The model also includes variables related to
human capital and income characteristics (of the household head) to control for differ-
ences in the dynamics followed toward the adoption decision. Geographic control of
distance (in log-Geodesic scale) to San Juan de Yapacaní, one of the leading centers
of rice technological diffusion, attempts to account for the level of exposure to the dis-
semination of rice technologies. Also, we use the farm’s size as a proxy for its wealth
endowment across all the decisions modeled, while yield priority represents households’
varietal preference for higher-yielding varieties. Finally, we included department-fixed
effects to account for the differences in labor availability and access to markets, since
evidence suggests that changes in these attributes may increase the opportunity cost
of technologies, making them less attractive or even nonaffordable (White, Labarta,
and Leguía 2005). Santa Cruz has the most extensive availability in both labor and mar-
kets; thus, we choose it as the base category in both regression analyses.
Descriptive statistics of all the explanatory variables used in the analysis are summa-
rized in Table 4, which compares average values between adopters and nonadopters of
the five complementary technologies used in rice production. The distribution of adop-
tion is somehow dissimilar across the different technological options. While roughly 65
percent of the rice growers use pesticide and weed controls, we found adoption levels of
MIVs and machinery slightly under 50 percent of the interviewed households. On the
other hand, fertilizer use is the lowest adopted rice technology (25 percent). Among
adopters, the household heads tend to be slightly more educated and younger than non-
adopters and, although female-headed households are unusual, they are marginally
fewer among adopters. Households adopting the rice technologies have, on average,
more alternative employment options and larger farm sizes. Adopters of technologies
seem to have more access to agricultural extension and farmers’organizations.
Statistical differences are significant between adopters and nonadopters, particularly
for the household head level of education, farm size, belonging to a farmers’association,
Agricultural and Resource Economics Review 9
https://doi.org/10.1017/age.2021.9
Downloaded from https://www.cambridge.org/core. IP address: 72.44.113.147, on 14 May 2021 at 15:02:45, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
Table 3. Description of variables
Variable Definition
MIVs (modern improved
varieties)
(1 = yes) Indicates whether the household uses modern improved
varieties in some of its plots. Modern varieties are set as those
disseminated on a large scale since 2004 (ten-year lag).
Machinery (1 = yes) Indicates whether the household uses large machinery
for planting or harvesting of rice plots.
Fertilizer (1 = yes) If the household uses mineral fertilizers for rice plots.
Pesticide (1 = yes) If the household uses chemical pesticides for rice plots.
Herbicide (1 = yes) If the household uses chemical herbicides for rice plots.
Yield priority (1 = yes) If the (expected) yield of a variety of rice is a priority
criterion for the farmer when choosing which variety to grow in
his/her field.
Expected sign: Positive since the search for more yield should
lead farmers to further adopt MIVs.
Education Years of formal education of the household’s head.
Expected relation: Positive, due to increased knowledge on
benefits from technology adoption.
Age Age (in years) of the household’s head.
Expected relation: Ambiguous; this happens because, although
it is related to more experience in crop management, it is also
related to more attachment to traditional technologies.
Female household head (1 = yes) If the household’s head is female.
Expected sign: Negative, related to a large portion of empirical
findings that demonstrate the less-favored conditions faced by
female-led farmer households.
Alternative employment (1 =yes) If at least one of the members of the household relies on
a different income source besides rice production.
Expected sign: Negative, if households depending solely on rice
production have a special interest in adopting better
technologies with respect to the rest.
Farm size In logarithmic form. Number of hectares available for agricultural
use.
Expected relation: Positive, because it is a proxy for farm/
household income with the literature suggesting that better-off
households are more likely to adopt modern technologies.
Association (1 = yes) If at least one household member belongs to a farmers’
association.
Expected sign: Positive, after considering feasible learning from
neighbors and other farmers.
Extension services (1 = yes) If, during the past two years, at least one household
member attended a workshop or extension service specializing
in rice farming.
Expected sign: Positive, due to improved knowledge on
management.
Distance to San Juan de
Yapacaní (SJY)
In logarithmic form. Geodesic distance from the farm’s location to
San Juan de Yapacaní cooperative CAISY.
Expected relation: Negative, since the more the distance to
improved technology dissemination and feasible spillover
effects, the less likely to adopt such technologies.
Source: Elaborated by the authors.
10 Martinez et al.
https://doi.org/10.1017/age.2021.9
Downloaded from https://www.cambridge.org/core. IP address: 72.44.113.147, on 14 May 2021 at 15:02:45, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
Table 4. Descriptive statistics
Technology MIVs Machinery Fertilizer Pest control Herbicide
Overall adoption rate 45.64% 49.13% 25.56% 65.84% 62.84%
Variables Overall
Mean
c
Nonadopter
a
Mean
Adopter
b
Mean
Nonadopter
Mean
Adopter
Mean
Nonadopter
Mean
Adopter
Mean
Nonadopter
Mean
Adopter
Mean
Nonadopter
Mean
Adopter
Mean
Yield priority
1
(%) 70.8 69.7 72.1 ————————
Education (years) 6.10 5.36 6.98*** 5.30 6.93*** 5.86 6.80** 5.51 6.40** 5.31 6.56***
Age ( years) 45.95 46.65 45.12* 46.24 45.66 46.23 45.14 48.14 44.82*** 47.04 45.31*
Females (%) 3.7 3.9 3.6 4.2 3.3 4.2 2.4 4.7 3.2 5.4 2.8*
Alt. employment (%) 48.6 46.6 51.1 49.0 48.2 47.6 51.7 49.3 48.3 48.0 49.0
Farm size (log) (ha) 2.93 2.55 3.38*** 2.12 3.76*** 2.64 3.76*** 2.09 3.36*** 1.87 3.55***
Association (%) 15.5 8.0 24.3*** 7.6 23.6*** 12.2 24.9*** 9.5 18.6*** 6.7 20.6***
Extension (%) 18.1 13.3 23.8*** 9.8 26.6*** 14.6 28.3*** 12.0 21.2*** 10.1 22.8***
Distance to SJY (log) 4.00 4.21 3.74*** 4.65 3.33*** 4.18 3.45*** 4.67 3.65*** 4.83 3.51***
(Log-) Distance to SJY (km) 90.1 106.37 72.28*** 128.68 51.6*** 103.31 54.4*** 136.2 67.27*** 145.9 58.24***
Beni (%) 28.8 35.8 20.5*** 46.8 10.2*** 35.3 9.8*** 50.7 17.4*** 57.7 11.7***
Cochabamba (%) 10.1 8.3 12.3* 13.2 6.9*** 11.4 6.3** 9.9 10.2 13.8 7.9***
Santa Cruz (%) 61.1 55.9 67.21*** 39.9 82.9*** 53.26 83.9*** 39.41 72.34*** 28.5 80.35***
1
Yield priority is disaggregated only for MIVs, following the rationale in Table 2.
a
Nonadopters,
b
adopters, and
c
reporting means and percentages for continuous and binary variables, respectively.
Reporting significance of variable’s mean difference between adopters and nonadopters of each technology: ***p < 0.01, **p < 0.05, *p < 0.1.
Source: Elaborated by the authors based on survey data.
Agricultural and Resource Economics Review 11
https://doi.org/10.1017/age.2021.9
Downloaded from https://www.cambridge.org/core. IP address: 72.44.113.147, on 14 May 2021 at 15:02:45, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
access to extension services, and the distance to the main technology diffusion center
(San Juan de Yapacaní). Notice how the difference of average distance to diffusion cen-
ters can be a scale factor of two when comparing adopters and nonadopters. In any
case, relying on alternative income sources is an evenly distributed condition between
adopters and nonadopters. On the other hand, having a female household head reports
only a statistical difference when comparing adopters and nonadopters of herbicides.
Results
Checking the sample unconditional and conditional probabilities of technology adop-
tion (Table 5), it is clear that there exists a strong complementarity among these five
practices. As an example, consider the unconditional probabilities of mechanization
(50.8 percent), fertilization (25.3 percent), use of pesticides (65.4 percent), and herbi-
cides (62.8 percent), that rapidly increase by 19.4, 11.3, 9.6, and 13.2 percent points,
respectively, given once the household adopts modern improved rice varieties. The sin-
gle condition that appears most increase other probabilities of adoption is inorganic fer-
tilizers, which raises the likelihood of using MIV to 65 percent, machinery to 87.4
percent, and both pesticide and herbicides to a solid 97.2 percent. Another interpreta-
tion is that farms that use fertilization (less recurrent practice) are more likely to be
those who have already implemented other technologies, hence use this final technology
to exploit their productive capabilities further. Notice how the probability of adoption
of other technologies systematically increases (sometimes over 95 percent) in any com-
bination of conditions that have fertilization as given. Also, the systematic increase in
the probability of adopting MIVs given the presence of a mechanized system (20 per-
cent point) may suggest that modern varieties are more common among farm house-
holds with higher initial endowments mechanization intensive. In general, without
further controls, these agricultural practices give a signal of being strongly, but not per-
fectly, correlated.
Our econometric analysis provides evidence that the decisions to adopt complemen-
tary technologies in rice production in Bolivia are correlated, justifying the use of an
MVP model to understand the factors that explain these decisions. Table 6 presents
the partial correlations across all technologies included in the analysis and the joint sig-
nificance test, suggesting that the individual adoption of rice technologies is affected by
the other complementary technologies’decisions. Not controlling for the correlation
between individual adoption decisions and their unobservables would bias the covari-
ates’coefficient estimation. When estimating separate Probit models for each technol-
ogy,
2
we find coefficients that are slightly larger in absolute value and more likely to be
statistically significant. Thus, assuming independent normal distributions for the errors
rather than a joint distribution may mean a potential distortion in the magnitude of the
detected partial correlations and an augmented probability of incurring in type I error
when defining the adoption determinants.
After controlling for the verified multivariate correlation of the decisions to adopt
complementary rice technologies and regional (department-level) fixed effects, we pre-
sent the results of the determinants of joint adoption in Table 7. We found that farm
size is the only explanatory variable that can explain the adoption of all the technologies
considered in the analysis. The greater the farm size, the higher the likelihood of
Bolivian rice growers adopting the available rice technologies. This finding is consistent
2
Not reported here. See Supplementary Table S4.
12 Martinez et al.
https://doi.org/10.1017/age.2021.9
Downloaded from https://www.cambridge.org/core. IP address: 72.44.113.147, on 14 May 2021 at 15:02:45, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
Table 5. Sample unconditional and conditional probabilities of technology adoption
Condition MIV Machinery Fertilization
Pest
control Herbicide
Unconditional 0.45 0.49 0.25 0.66 0.63
Y
V
= 1 1 0.7*** 0.37*** 0.75*** 0.76***
Y
M
= 1 0.65*** 1 0.44*** 0.85*** 0.93***
Y
F
= 1 0.68*** 0.87*** 1 0.97*** 0.97***
Y
P
= 1 0.54*** 0.66*** 0.38*** 1 0.81***
Y
H
= 1 0.57*** 0.75*** 0.39*** 0.85*** 1
Y
V
=1, Y
M
= 1 1 1 0.48*** 0.88*** 0.95***
Y
V
=1, Y
F
= 1 1 0.92*** 1 0.98*** 0.97***
Y
V
=1, Y
P
= 1 1 0.82*** 0.48*** 1 0.9***
Y
V
=1, Y
H
= 1 1 0.88*** 0.47*** 0.88*** 1
Y
M
=1, Y
F
= 1 0.72*** 1 1 0.98*** 0.99***
Y
M
=1, Y
P
= 1 0.67*** 1 0.5*** 1 0.97***
Y
M
=1, Y
H
= 1 0.66*** 1 0.46*** 0.88*** 1
Y
F
=1, Y
P
= 1 0.68*** 0.88*** 1 1 0.98***
Y
F
=1, Y
H
= 1 0.68*** 0.89*** 1 0.98*** 1
Y
P
=1, Y
H
= 1 0.59*** 0.78*** 0.45*** 1 1
Y
V
=1, Y
M
=1, Y
F
= 1 1 1 1 0.99*** 0.99***
Y
V
=1, Y
M
=1, Y
P
= 1 1 1 0.54*** 1 0.97***
Y
V
=1, Y
M
=1, Y
H
= 1 1 1 0.5*** 0.89*** 1
Y
V
=1, Y
F
=1, Y
P
= 1 1 0.93*** 1 1 0.98***
Y
V
=1, Y
F
=1, Y
H
= 1 1 0.94*** 1 0.99*** 1
Y
V
=1, Y
P
=1, Y
H
= 1 1 0.89*** 0.52*** 1 1
Y
M
=1, Y
F
=1, Y
P
= 1 0.72*** 1 1 1 0.99***
Y
M
=1, Y
F
=1, Y
H
= 1 0.71*** 1 1 0.98*** 1
Y
M
=1, Y
P
=1, Y
H
= 1 0.67*** 1 0.51*** 1 1
Y
F
=1, Y
P
=1, Y
H
= 1 0.68*** 0.89*** 1 1 1
Y
V
=1, Y
M
=1, Y
F
=1,
Y
P
=1
1 1 1 1 0.98***
Y
V
=1, Y
M
=1, Y
F
=1,
Y
H
=1
1 1 1 0.98*** 1
Y
V
=1, Y
M
=1, Y
P
=1,
Y
H
=1
1 1 0.55*** 1 1
Y
V
=1, Y
F
=1, Y
P
=1,
Y
H
=1
1 0.94*** 1 1 1
(Continued)
Agricultural and Resource Economics Review 13
https://doi.org/10.1017/age.2021.9
Downloaded from https://www.cambridge.org/core. IP address: 72.44.113.147, on 14 May 2021 at 15:02:45, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
with other studies implemented in Latin America’s rice sector (Suarez 2014; Marin et al.
2018) and further studies on agricultural technology adoption more broadly in different
regions (Walker and Alwang 2015; Yamano et al. 2016). A plausible explanation of this
result is that larger farmers tend to have better conditions and more resources to cope
with the risks of incorporating new technology in their current crop management. In
Bolivia, it has been reported that farmers with better socioeconomic conditions are
more likely to have better practices in their agricultural systems (Ortiz and Soliz
2007; Salazar et al. 2015).
Two other factors were relevant in explaining farmers’decisions to adopt three out of
the five rice technologies. Being part of a farmer association significantly increases the
probability of adopting MIVs, machinery, and herbicides. This finding can be explained
by the fact that most programs delivering agricultural inputs in Bolivia use existing
organizations for reaching and distribution purposes. Likewise, having access to agricul-
tural extension services was found to increase the probability of adopting machinery
and fertilizer. Although extension services have had limited coverage in Bolivia, both
public and NGO extension programs have tried to meet local farmers’demands and
make available these agricultural technologies (FAO 2011). This has been a strong argu-
ment for organizations such as the Centro de Investigación Agrícola Tropical in design-
ing new strategies to overcome current restrictions on farmers in specific regions to
access known rice technologies.
As expected, we found a negative relationship between distance to the main center of
rice technological dissemination in San Juan de Yapacaní and the probability of
Table 5. (Continued.)
Condition MIV Machinery Fertilization Pest
control
Herbicide
Y
M
=1, Y
F
=1, Y
P
=1,
Y
H
=1
0.72*** 1 1 1 1
Y
V
,Y
M
,Y
F
,Y
P
, and Y
H
are index variables for the adoption status (1 = yes) for modern improved varieties, machinery,
fertilization, pest control, and herbicides, respectively.
***p < 0.01, **p < 0.05, *p < 0.1, compared with the unconditional probability of adopting the technology.
Source: Elaborated by the authors based on survey data.
Table 6. Multivariate Probit correlation results
ρ
21
ρ
31
ρ
41
ρ
51
ρ
32
0.463*** 0.238*** 0.140 0.204* 0.609***
(0.10) (0.08) (0.10) (0.10) (0.09)
ρ
42
ρ
52
ρ
43
ρ
53
ρ
54
0.426*** 0.909*** 0.776*** 0.704*** 0.623***
(0.09) (0.12) (0.11) (0.11) (0.10)
Joint significance LR test: chi
2
(10) = 379.43, p-value = 0.00.
Robust standard errors in parentheses. ***p < 0.01, **p < 0.05, *p < 0.1.
Note: ρ
ij
defines the correlation between technologies iand j, with values meaning (1) MIVs, (2) machinery, (3) fertilizer,
(4) pesticide, and (5) herbicide.
Source: Elaborated by the authors based on survey data.
14 Martinez et al.
https://doi.org/10.1017/age.2021.9
Downloaded from https://www.cambridge.org/core. IP address: 72.44.113.147, on 14 May 2021 at 15:02:45, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
Table 7. Multivariate Probit regression results on technology adoption
(1) (2) (3) (4) (5)
Variables MIVs Machinery Fertilizer Pesticide Herbicide
Yield as first interest when choosing a variety 0.091 (0.09)
Years of education of household head 0.043*** (0.01) 0.057*** (0.01) 0.010 (0.02) 0.001 (0.01) 0.028 (0.02)
Age (years) of household head 0.000 (0.00) −0.000 (0.01) −0.002 (0.00) −0.016*** (0.00) −0.009* (0.01)
Female household head −0.070 (0.21) −0.343 (0.25) −0.393 (0.26) −0.406* (0.22) −0.859*** (0.25)
Alt. income-generating employment 0.023 (0.10) −0.107 (0.09) 0.080 (0.10) −0.014 (0.10) 0.061 (0.12)
(Log-) Size of the farm 0.109* (0.06) 0.190*** (0.05) 0.149*** (0.04) 0.129*** (0.04) 0.138*** (0.05)
Part of production association 0.594*** (0.22) 0.564* (0.30) 0.275 (0.21) 0.228 (0.27) 0.654* (0.37)
Access to extension services 0.126 (0.18) 0.351** (0.15) 0.386*** (0.13) 0.108 (0.15) 0.186 (0.16)
Fixed effect
a
: Beni −0.105 (0.26) −0.064 (0.36) −0.673*** (0.25) −0.106 (0.28) −0.331 (0.41)
Fixed effect
a
: Cochabamba 0.260 (0.28) −0.463 (0.31) −0.374** (0.19) −0.113 (0.34) −0.794** (0.33)
(Log-) Distance to San Juan de Yapacaní −0.069 (0.07) −0.726*** (0.18) −0.065 (0.07) −0.432*** (0.15) −0.813*** (0.22)
Constant −0.619 (0.39) 2.042*** (0.69) −0.846** (0.40) 2.638*** (0.63) 3.786*** (0.89)
Observations 802 802 802 802 802
a
Base category is Santa Cruz.
Robust standard errors in parentheses. ***p < 0.01, **p < 0.05, *p < 0.1.
Source: Elaborated by the authors based on survey data.
Agricultural and Resource Economics Review 15
https://doi.org/10.1017/age.2021.9
Downloaded from https://www.cambridge.org/core. IP address: 72.44.113.147, on 14 May 2021 at 15:02:45, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
adopting rice technologies; however, we found significant effects only on adopting
machinery, pesticides, and herbicides. This hypothesis, first developed by Amemiya
(2001), has been largely confirmed in the literature. Furthermore, Taboada et al.
(2005) has highlighted the relevance of the Japanese-Bolivian dissemination centers
for Bolivia’s rice production. For the case of adopting MIVs and fertilizer, it could
be argued that these specialized technologies may show asymmetries due to geographic
limitations. The latter could be because those farm households near Japanese-Bolivian
farmers observe the technological measures taken by those farmers, which include rel-
atively high-cost inputs such as machinery and chemical pesticides and herbicides.
Nonetheless, recent studies among rural communities in Bolivia show that informa-
tion technology may substantially enhance rural livelihoods (Gigler 2015) by helping
traditionally marginalized communities make better-informed decisions. However,
such impacts may heavily rely on strong intermediate actors (in our case, extension ser-
vices and associations), who can further help farmers assimilate technical information
and recommendations provided by research entities. Despite such findings, increasing
availability of information technologies in remote areas means a need for future studies
with updated data that test whether the effect of proximity to dissemination centers still
holds.
Other factors that showed significant effects on rice technologies’adoption decisions
are the household head’s age, education level, and gender. An additional year in the
household head’s education significantly affects the probability of adopting MIVs and
machinery. Likewise, we found that older farm households and female-headed house-
holds are less likely to adopt chemical pest and weed controls. These findings are con-
sistent with previous results reported in rice in Africa (Saka et al. 2005; Usman, Ango,
and Barau 2013) and Latin America (Suarez 2014), as well as for other crops and prac-
tices in developing countries (Kassie et al. 2013; Teklewold, Kassie, and Shiferaw 2013;
Manda et al. 2016; Gebremariam and Tesfaye, 2018; Kanyenji et al. 2020).
We summarize the results from the ordered Probit model of the number of technol-
ogies adopted by Bolivian rice farmers in Table 8. One of the first striking results is
the systematic change of sign for all marginal effects when crossing the barrier of
two or fewer technologies to three or more technologies. On average, farms whose
household heads are more educated are slightly over 1 percent more likely to adopt
three or more technologies. Also, larger farms (i.e., related to higher financial endow-
ments) are about 8 percent more likely to adopt more technologies. Also, farming
households who are part of a production association or have access to extension services
are 20 and 12 percent more likely, respectively, to be among those with higher levels of
adoption.
On the other hand, farms that are farther away from the national dissemination cen-
ter are around 8 percent more likely to adopt two or fewer technologies. Furthermore,
households located in Beni and Cochabamba will have about 23 and 13 percent
more probability, respectively, to find themselves in the lower range of technology
adoption. Also, farms whose household head is female are 17 percent less likely to
reach upper adoption levels. Off-farm means of income seem to have no effect on
farmers being either in the upper or lower adoption levels. The only apparent inconsis-
tent result is the null effect of the household head’s age, but this may be a consequence
of the small (although significant) correlation it holds with only two technologies,
namely pesticides and herbicides. These additional findings corroborate the determi-
nants highlighted in the multivariate model and extend them to the total amount of
adoption.
16 Martinez et al.
https://doi.org/10.1017/age.2021.9
Downloaded from https://www.cambridge.org/core. IP address: 72.44.113.147, on 14 May 2021 at 15:02:45, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
Table 8. Ordered Probit regression and marginal effects on the number of adopted technologies
(1) (2) (3) (4) (5) (6) (7)
Variables Coeff. P(k=0) P(k=1) P(k=2) P(k=3) P(k=4) P(k=5)
Years of educ. of household
head
0.03*** (0.01) −0.01*** (0.00) −0.01*** (0.00) −0.00*** (0.00) 0.00*** (0.00) 0.01*** (0.00) 0.00*** (0.00)
Age (years) of household head −0.00 (0.00) 0.00 (0.00) 0.00 (0.00) 0.00 (0.00) −0.00 (0.00) −0.00 (0.00) −0.00 (0.00)
Female household head −0.44** (0.20) 0.09* (0.05) 0.07*** (0.03) 0.00 (0.01) −0.04* (0.02) −0.08** (0.03) −0.05*** (0.02)
Alt. income-generating
employment
0.02 (0.08) −0.00 (0.01) −0.00 (0.01) −0.00 (0.00) 0.00 (0.00) 0.00 (0.01) 0.00 (0.01)
(Log-) Size of the farm 0.19*** (0.03) −0.03*** (0.01) −0.04*** (0.01) −0.01*** (0.00) 0.01*** (0.00) 0.04*** (0.01) 0.03*** (0.00)
Part of production association 0.51*** (0.11) −0.07*** (0.01) −0.09*** (0.02) −0.04*** (0.01) 0.01* (0.01) 0.09*** (0.02) 0.10*** (0.03)
Access to extension services 0.29*** (0.11) −0.04*** (0.01) −0.06*** (0.02) −0.02** (0.01) 0.01*** (0.00) 0.05*** (0.02) 0.05** (0.02)
Fixed effect
a
: Beni −0.61*** (0.12) 0.12*** (0.03) 0.10*** (0.02) 0.01*** (0.01) −0.05*** (0.01) −0.11*** (0.02) −0.08*** (0.01)
Fixed effect
a
: Cochabamba −0.32** (0.13) 0.06** (0.03) 0.06*** (0.02) 0.01*** (0.00) −0.03* (0.01) −0.06** (0.02) −0.04*** (0.01)
(Log-) Distance to San Juan de
Yapacaní
−0.22*** (0.04) 0.04*** (0.01) 0.04*** (0.01) 0.01*** (0.00) −0.01*** (0.00) −0.04*** (0.01) −0.03*** (0.01)
λ
1
−1.75*** (0.25)
λ
2
−0.92*** (0.25)
λ
3
−0.42* (0.25)
λ
4
0.13 (0.25)
λ
5
0.96*** (0.25)
Observations 802 802 802 802 802 802 802
a
Base category is Santa Cruz.
Joint significance LR test: chi
2
(10) = 394.33, p-value = 0.00.
Robust standard errors in parentheses. ***p < 0.01, **p < 0.05, *p < 0.1.
Source: Elaborated by the authors based on survey data.
Agricultural and Resource Economics Review 17
https://doi.org/10.1017/age.2021.9
Downloaded from https://www.cambridge.org/core. IP address: 72.44.113.147, on 14 May 2021 at 15:02:45, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
Conclusions
Despite the importance of rice in Bolivia, this cereal crop production remains constrained
and shows low yields. Our results show that Bolivian rice growers have adopted most of
the five agricultural technologies studied (MIVs, machinery, fertilizer, pesticide, and
herbicide), although at different rates. However, this adoption has not translated into
higher rice yields that better position Bolivia’s productivity among other LAC countries.
Our results reveal strong complementarities among rice-producing technologies in
Bolivia, which must be considered when exploring their adoption determinants. This
further reinforces previous findings that highlight the caveats of assuming independent
decisions on productive technologies adoption—it may result in biased estimates and
increased likelihood of type I error at identifying specific determinants of adoption.
Modern varieties have reached a rate of adoption of 45.6 percent, which is a good
outcome for the nascent rice breeding program in the country. The materials MAC
18 and IAC 101 are the most used ones because of their productivity and good grain
quality. Yet again, the adoption of these varieties seems to have not influenced rice pro-
ductivity. This could be associated with the relatively low use of fertilizer (25 percent), tra-
ditionally referred to as a perfect complement of improved varieties. Although MIVs are
suitable for manual and mechanized systems, their adoption appears to be more concen-
trated among mechanized systems. Hence, the potential for improvement for the overall
rice productive system in Bolivia may have not yet been reached.
Our analysis particularly highlights the critical role of belonging to a farmer organi-
zation, having access to agricultural extension, and proximity to San Juan de Yapacaní
(the main technological dissemination area for rice in Bolivia), in increasing the likeli-
hood of adopting improved technologies in rice production. These determinants hold
not only for the adoption decision of each technology but also for the extent of adop-
tion of these technologies among Bolivian rice farmers. Farms located in areas with
higher availability of labor and large markets (namely, the Department of Santa
Cruz) are more likely to adopt a larger number of technologies.
Therefore, the weakness of the national extension system can also explain part of the
low rice yields despite the documented adoption of rice technologies. Our results also
support strategies that try to build technology dissemination efforts on the country’s
existing strengths. Bolivia is a country that has traditionally relied on local farmer orga-
nizations, and these should continue to be a critical mechanism for further technology
dissemination. Likewise, government policy should further promote and expand the use
of existing technology diffusion centers such as San Juan de Yapacaní. This has proven
to be an efficient farmer-to-farmer knowledge dissemination mechanism, especially
among small and medium rice growers for whom low rice yields are concentrated.
Finally, this study also provides evidence that technology adoption in developing countries
seems biased toward larger farmers, who respond better to using existing improved
technologies. Governments and donors should concentrate efforts on identifying strategies
to target resource-poor farmers better, so they can gain access to improved technologies,
which may result in an overall increase in rice productivity and livelihood conditions.
Supplementary material. The supplementary material for this article can be found at https://doi.org/10.
1017/age.2021.9.
Acknowledgments. We want to thank Victor Zuluaga, Alexander Buriticá, Javier Castro, and the 8th
Bolivian Conference of Development Economics participants for the comments provided to earlier versions
of the manuscript. Also, we thank the comments and suggestions from two anonymous referees and the
Editor, Dr. Richard Melstrom, which greatly benefited the article.
18 Martinez et al.
https://doi.org/10.1017/age.2021.9
Downloaded from https://www.cambridge.org/core. IP address: 72.44.113.147, on 14 May 2021 at 15:02:45, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
Funding statement. This study is a result of the project Adoption study of rice varieties in Bolivia led by
the Alliance of Bioversity International and CIAT, made possible with support from the CGIAR Global
Research Program on Rice and HarvestPlus. HarvestPlus’principal donors are the UK Government; the
Bill & Melinda Gates Foundation; the US Government’s Feed the Future initiative; Government of
Canada; the European Commission; and donors to the CGIAR Research Program on Agriculture for
Nutrition and Health (A4NH). HarvestPlus is also supported by the John D. and Catherine
T. MacArthur Foundation.
Conflicts of interest. None.
References
Amemiya, K. 2001. “The Importance of Being Japanese in Bolivia”. Japan Policy Research Institute
Working Paper 75.
Andersen, L., and D. Verner. 2009. “Social Impacts of Climate Change in Bolivia: A Municipal Level
Analysis of the Effects of Recent Climate Change on Life Expectancy, Consumption, Poverty and
Inequality”. World Bank Working Paper 5092.
Aryal, J.P., D.B. Rahut, S. Maharjan, and O. Erenstein. 2018. “Factors Affecting the Adoption of Multiple
Climate-Smart Agricultural Practices in the Indo-Gangetic Plains of India.”Natural Resources Forum 42
(1): 141–158.
Bardhan, P., and C. Udry. 1999. Development Microeconomics. Oxford, UK: Oxford University Press.
Boersch-Supan, A., and V. Hajivassiliou. 1990. “Smooth Unbiased Multivariate Probability Simulators for
Maximum Likelihood Estimation of Limited Dependent Variable Models.”Journal of Econometrics 58
(3): 347–368.
Calpe, C. 2006. Rice International Commodity Profile. December 2006. Rome, Italy: Food and Agriculture
Organization of the United Nations.
Cameron, A.C., and P.K. Trivedi. 2005. Microeconometrics: methods and applications.Oxford,UK:Oxford
University Press.
Châtel, M., E.P. Guimarâes, Y. Ospina, F. Rodríguez, and V.H. Lozano. 2010. “Mejoramiento de pobla-
ciones de arroz de secano empleando selección recurrente y desarrollo de variedades.”In: V. Degiovani,
C.P. Martínez and F. Motta (eds.), Producción Eco-Eficiente del Arroz en América Latina, Tomo I (pp.
191–206). Cali, Colombia: International Center for Tropical Agriculture.
Dalrymple, D.G. 1979. “The Adoption of High-yielding Varieties in Developing Nations.”Agricultural
History 53(4): 704–726.
Dalrymple, D.G. 1986. Development and Spread of High-Yielding Rice Varieties in Developing Countries.
Washington, D.C.: Agency for International Development.
Donkoh, S.A., S.B. Azumah, and J.A. Awuni. 2019. “Adoption of Improved Agricultural Technologies
among Rice Farmers in Ghana: A Multivariate Probit Approach.”Ghana Journal of Development
Studies 16(1): 47–67.
FAO. 2011. “Communication for Agricultural Innovation in Bolivia: The Challenge of Institutionalization.
Case Study.”Rome: Communication for Sustainable Development Initiative (CSDI). 93 pp.
FAOSTAT. 2020. “Food and Agriculture Organization of the United Nations.”Available at http://www.fao.
org/faostat/en/.
Feder, G., R. Just, and D. Zilberman. 1985. “Adoption of Agricultural Innovations in Developing
Countries: A Survey.”Economic Development and Agricultural Change 33(2): 255–298.
Foster, D.A., and M. Rosenzweig. 2010. “Microeconomics of Technology Adoption.”Annual Review of
Economics 2(1): 395–424.
Gebremariam, G., and W. Tesfaye. 2018. “The Heterogeneous Effect of Shocks on Agricultural
Innovations Adoption: Microeconometric Evidence from Rural Ethiopia.”Food Policy 74(1): 154–161.
Gigler, B.S. 2015. Development as Freedom in a Digital Age: Experiences from the Rural Poor in Bolivia.
Washington, DC: The World Bank.
Godoy, R., J. Morduch, and D. Bravo. 1998. “Technological Adoption in Rural Cochabamba, Bolivia.”
Journal of Anthropological Research 54(3): 351–372.
Greene, W.H. 2012. Econometric Analysis, 7th Edition. Upper Saddle River, NJ: Prentice-Hall.
Agricultural and Resource Economics Review 19
https://doi.org/10.1017/age.2021.9
Downloaded from https://www.cambridge.org/core. IP address: 72.44.113.147, on 14 May 2021 at 15:02:45, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
GRiSP (Global Rice Science Partnership). 2013. Rice Almanac, 4th Edition. Los Baños, Philippines:
International Rice Research Institute.
HarvestPlus. 2009. “Bolivia lanzó comercialmente semillas de arroz con mayores nutrientes.”Available at
http://lac.harvestplus.org/bolivia-lanzo-comercialmente-semillas-de-arroz-con-mayores-nutrientes/.
Hoffman, V., K. Probst, and A. Christinck. 2007. “Farmers and Researchers: How Can Collaborative
Advantages Be Created in Participatory Research and Technology Development?”Agriculture and
Human Values 24(1): 355–368.
Kanyenji, G.M., W. Oluoch-Kosura, C.M. Onyango, and S.K. Ng’ang’a. 2020. “Prospects and Constraints
in Smallholder Farmers’Adoption of Multiple Soilcarbon Enhancing Practices in Western Kenya.”
Heliyon 6(3): e03226, 1–10.
Kassie, M., M. Jaleta, B. Shiferaw, F. Mmbando, and M. Mekuria. 2013. “Adoption of Interrelated
Sustainable Agricultural Practices in Smallholder Systems: Evidence from Rural Tanzania.”
Technological Forecasting & Social Change 80(3): 525–540.
Labarta, R.A., A. Buriticá, D.C. Lopera, C. González, M. Del Río, S. Pérez, and R. Andrade. 2014.
“Contribución del CIAT al mejoramiento genético de arroz en 7 países de América Latina.”CIAT
Working Paper.
Lynch, J., and E. Tasch. 1981. “Programa del CIAT para la investigación en Arroz de Secano en América
Latina.”Paper presented at the Seminar Estrategia de investigación y políticas agrícolas, January 14–16.
International Center for Tropical Agriculture (CIAT).
Manda, J., A.D. Alene, C. Gardebroek, M. Kassie, and G. Tembo. 2016. “Adoption and Impacts of
Sustainable Agricultural Practices on Maize Yields and Incomes: Evidence from Rural Zambia.”
Journal of Agricultural Economics 67(1): 130–153.
Marin, D., R. Andrade, R.A. Labarta, and J. Twyman. 2018. “Participación de la mujer en las decisiones
sobre el uso y la intensidad de siembra de variedades de arroz en Ecuador.”Cuestiones Económicas,
Número especial de Economía y Género 28(3): 119–146.
Mponella, P., G.T. Kassie, and L.D. Tamene. 2018. “Simultaneous Adoption of Integrated Soil Fertility
Management Technologies in the Chinyanja Triangle, Southern Africa.”Natural Resources Forum
42(1): 172–184.
Nguyen, V.N., and D.V. Tran. 2002. “Rice in Producing Countries.”In Food and Agriculture Organization
of the United Nations (ed.), FAO Rice Information, vol. 3.
Oladimeji, T.E., O. Oyinbo, A.A. Hassan, and O. Yusuf. 2020. “Understanding the Interdependence and
Temporal Dynamics of Smallholders’Adoption of Soil Conservation Practices: Evidence from Nigeria.”
Sustainability 12(7): 2376, 1–21.
Ortiz, A., and L. Soliz. 2007. El arroz en Bolivia. La Paz, Bolivia: Centro de Investigación y Promoción del
Campesinado (CIPCA).
Pretty, J., R. Ruben, and L.A. Thrupp. 2002. “Institutional Changes and Policy Reforms.”In: E. Uphoff
(ed.), Agroecological Innovations: Increasing Food Production with Participatory Development (pp. 251–
260). London, UK: Earthscan Publishing Ltd.
Saka, J.O., V.O. Okoruwa, B.O. Lawal, and S. Ajijola. 2005. “Adoption of Improved Rice Varieties
among Small-Holder Farmers in Southwestern Nigeria.”World Journal of Agricultural Sciences 1(1):
42–49.
Salazar, L., J. Aramburu, M. Gonzalez-Flores, and P. Enters. 2015. “Food Security and Productivity:
Impacts of Technology Adoption on Small Subsistence Farmers in Bolivia.”Inter-American
Development Bank. Working Paper Series IDB-WP-567.
Suarez, S. 2014. “Determinantes de adopción e intensidad de adopción de variedades mejoradas modernas
de arroz en el norte de Perú”Bachelor thesis. Economics Department, Universidad del Valle, Colombia.
Taboada, R., R. Guzman, J. Viruez, and T. Kon. 2005. “Rice Population Improvement in Bolivia.”In E.
Guimarães (ed.), Population Improvement: A Way of Exploiting the Rice Genetic Resources of Latin
America (pp. 221–236). Rome, Italy: Food and Agriculture Organization of the United Nations.
Teklewold, H., M. Kassie, and B. Shiferaw. 2013. “Adoption of Multiple Sustainable Agricultural Practices
in Rural Ethiopia.”Journal of Agricultural Economics 64(3): 597–623.
Usman, T., A. Ango, and A. Barau. 2013. “Evaluation of Adoption of Improved Rice Varieties Among
Small-Scale Farmers: A Case of Goronyo Local Government Area of Sokoto State, North-Western
Nigeria.”International Journal of Agricultural Innovations and Research 2(3): 408–414.
20 Martinez et al.
https://doi.org/10.1017/age.2021.9
Downloaded from https://www.cambridge.org/core. IP address: 72.44.113.147, on 14 May 2021 at 15:02:45, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
Viruez, J., and R. Taboada. 2013. “Producción de arroz en Bolivia: conocimiento técnico para un manejo
eficiente y rentable.”Centro de Investigación Agrícola Tropical and Federación Nacional de Cooperativas
Arroceras (FENCA).
Wainaina, P., S. Tongruksawattanab, and M. Qaima. 2016. “Tradeoffs and Complementarities in the
Adoption of Improved Seeds, Fertilizer, and Natural Resource Management Technologies in Kenya.”
Agricultural Economics 47(1): 351–362.
Walker, T.S., and J. Alwang. 2015. “Crop Improvement, Adoption, and Impact of Improved Varieties in
Food Crops in Sub-Saharan Africa.”Montpellier, France: CAB International. CGIAR Consortium of
International Agricultural Research Centers.
Ward, P.S., A.R. Bellb, K. Droppelmannc, and T.G. Bentond. 2018. “Early Adoption of Conservation
Agriculture Practices: Understanding Partial Compliance in Programs with Multiple Adoption
Decisions.”Land Use Policy 70(1): 27–37.
White, D.S., R.A. Labarta, and E.J. Leguía. 2005. “Technology Adoption by Resource-Poor Farmers:
Considering the Implications of Peak-Season Labor Costs.”Agricultural Systems 85(2): 183–201.
Winters, C. 2012. “Impact of Climate Change on the Poor in Bolivia.”Global Majority E-Journal 3(1): 33–43.
Wollni, M., D.R. Lee, and J.E. Thies. 2010. “Conservation Agriculture, Organic Marketing, and Collective
Action in the Honduran Hillsides.”Agricultural Economics 41(3–4): 373–384.
Wu, JunJie, and Bruce A Babcock. 1998. “The Choice of Tillage, Rotation, and Soil Testing Practices:
Economic and Environmental Implications.”American Journal of Agricultural Economics 80(3): 494–
511.
Yamano, T., A. Arouna, R.A. Labarta, Z.M. Huelgas, and S. Mohanty. 2016. “Adoption and Impacts of
International Rice Research Technologies.”Global Food Security 8(1): 1–8.
Yirga, C., Y. Atnafe, and A. AwHassan. 2015. “A Multivariate Analysis of Factors Affecting Adoption of
Improved Varieties of Multiple Crops: A Case Study from Ethiopian Highlands.”Ethiopian Journal of
Agricultural Sciences 25(2): 29–45.
Zeng, Y., Y. Tian, K. He, and J. Zhang. 2020. “Environmental Conscience, External Incentives and Social
Norms in Rice Farmers’Adoption of Pro-environmental Agricultural Practices in Rural Hubei Province,
China.”Environmental Technology 41(19): 2518–2532.
Cite this article: Martinez JM, Labarta RA, Gonzalez C, Lopera DC (2021). Joint adoption of rice technol-
ogies among Bolivian farmers. Agricultural and Resource Economics Review 1–21. https://doi.org/10.1017/
age.2021.9
Agricultural and Resource Economics Review 21
https://doi.org/10.1017/age.2021.9
Downloaded from https://www.cambridge.org/core. IP address: 72.44.113.147, on 14 May 2021 at 15:02:45, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
