Agricultural Economics 26 (2001) 267–280
Factors explaining the diffusion of hybrid maize in
Latin America and the Caribbean region
Jennifer L. Kosarek a,b,1 , Philip Garcia a,∗ , Michael L. Morris b,2
a
b
Department of Agricultural and Consumer Economics, University of Illinois at Urbana-Champaign,
1301 W. Gregory Drive, Urbana, IL 61801, USA
Economics Program, International Maize and Wheat Improvement Center (CIMMYT), Apdo. Postal 6-641, Mexico, D.F. 06600, Mexico
Received 8 February 2000; received in revised form 29 September 2000; accepted 25 October 2000
Abstract
If future demand for maize in Latin America and the Caribbean region (LAC) is to be met from local sources, domestic
production must continue to increase. Because further expansion in the area planted to maize is precluded by the limited
availability of arable land, future increases in production will have to rely heavily on the spread of productivity-enhancing
hybrid technology. Until now, the diffusion of hybrid maize in LAC has been quite variable. Using data from 18 countries, we
investigate factors affecting the hybrid maize diffusion rate. Our findings validate conventional profitability-based explanations
of producer adoption behavior, but they also confirm the importance of supply-side factors, thereby providing empirical support
for the life cycle theory of seed industry development. We conclude that if policy makers in LAC are to accelerate the diffusion
of hybrid maize, they will have to ensure an environment in which it is not only profitable for producers to adopt improved
germplasm, but also profitable for the seed industry to produce and sell high-quality seed. © 2001 Elsevier Science B.V. All
rights reserved.
JEL classification: Q16; N56
Keywords: Maize; Diffusion; Latin America; Caribbean
1. Introduction
In terms of global production, maize is the third
most important cereal after wheat and rice. In Latin
America and the Caribbean region (LAC), where it is
currently planted on over 26 million hectares (FAO,
2000), maize is the leading food staple, the principal
ingredient in most livestock feeds, and a mainstay of
∗ Corresponding author. Tel.: +1-217-333-0644;
fax: +1-217-333-5538.
E-mail addresses: [email protected] (J.L. Kosarek),
[email protected] (P. Garcia), [email protected] (M.L. Morris).
1 Tel.: +1-217-333-3209; fax: +1-217-333-5538.
2 Tel.: +1-52-5804-2004; fax: +1-52-5804-7558.
the rural economy. According to projections made by
the International Food Policy Research Institute, during the period 1990–2020 demand for maize in LAC
will grow at an average annual rate of over 2%, fueled by population growth and per capita income gains
(Rosegrant et al., 1995). If maize supply in LAC is to
keep pace with this projected growth in demand, domestic production will have to increase significantly.
Production increases in maize can be achieved
through expansion in area planted, yield gains, or
some combination of the two. Diminishing availability
of arable land in LAC rules out the possibility of further expansion in area planted, suggesting that future
production growth will depend mainly on yield gains
0169-5150/01/$ – see front matter © 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 1 6 9 - 5 1 5 0 ( 0 0 ) 0 0 1 2 9 - 8
268
J.L. Kosarek et al. / Agricultural Economics 26 (2001) 267–280
Fig. 1. Diffusion of hybrid maize, Latin America and Caribbean region, 1996.
made possible by the spread of productivity-enhancing
technologies. One particularly important source of potential productivity gains is the improved germplasm
contained in hybrid maize varieties. 3
Currently, use of hybrid maize varies considerably
throughout LAC. In a few countries (e.g., Venezuela,
Argentina), a relatively high proportion of the national
maize area is planted to hybrids, but adoption has been
much less extensive in many other countries, including
several in which use of hybrids is practically unknown
(e.g., Haiti, Dominican Republic, Nicaragua). Across
the region as a whole, over 50% of the total maize
area is still planted to farmer’s traditional varieties
(Fig. 1). Thus, substantial productivity gains could be
achieved by introducing hybrid maize technology into
3
Until recently, hybrids were considered appropriate only for
large-scale commercial producers located in favorable production environments. Evidence is accumulating to suggest that
well-adapted hybrids can perform well even in marginal environments and under reduced levels of management, conditions that
often characterize smallholder agriculture in developing countries
(Heisey et al., 1998).
areas in which local varieties continue to dominate.
In this context, identifying factors associated with the
successful diffusion of hybrids may be useful to policy
makers interested in developing strategies to increase
maize production.
Technology adoption decisions in developing countries have been extensively analyzed (for surveys of
the adoption literature, see Feder et al., 1985; Rauniyar and Goode, 1992). Complementing the large
amount of theoretical work that focuses on technology
adoption in general, numerous empirical case studies
provide a wealth of information about the factors affecting the farm-level decision to adopt hybrid maize
(e.g., see Gerhart, 1975; Walker, 1981; CIMMYT,
1992; Byerlee et al., 1993; Smale et al., 1991, 1995;
Kumar, 1994; Heisey et al., 1998). The common
theme emerging from this literature is that the decision to adopt hybrid maize is influenced by a complex
and highly variable set of factors. Depending on the
context, these can include demographic characteristics
of the household (e.g., size, age and gender composition, wealth, education level of the household head),
J.L. Kosarek et al. / Agricultural Economics 26 (2001) 267–280
the expected profitability and/or perceived riskiness
of the technology, farmers’ consumption preferences,
and the availability and cost of inputs, especially seed.
While the farm-level decision to adopt hybrid maize
has been the focus of considerable research, much less
work has been done at the aggregate industry level to
identify factors that influence the diffusion of hybrid
technology. In his pioneering study of the spread of
hybrid maize in the United States, Griliches (1957)
hypothesized that the uneven rate of diffusion could
be linked to both demand and supply factors. Griliches
determined that variability in the demand for hybrid
maize is related to the additional profits that farmers
expect to gain by switching from open-pollinated varieties (OPV) to hybrids. He also found that variability
in the supply of hybrid seed is related to the revenue
that seed suppliers expect to earn by entering the market, which depends on factors such as the size of the
market, marketing costs, product innovation costs, and
expected rate of acceptance.
More recently, Heisey et al. (1998) used crosssectional data to investigate how demand and supply
factors influence the spread of hybrid maize in 32
developing countries. Heisey et al. concluded that
at the aggregate (country) level, diffusion of hybrid
maize depends partly on the expected profitability of
the technology, which is driven by germplasm performance and seed price. As well, they determined
that the diffusion of hybrid maize is strongly influenced by industry-level profitability, which depends
on characteristics of the seed market, the organization
of the local seed industry, and the cost of research
innovation, among other factors.
The empirical work by Griliches and Heisey et al.
highlights an important but frequently overlooked
point: even if farmers determine that they can potentially benefit by adopting hybrids, without adequate
supply of hybrid seed, successful adoption cannot
occur. In attempting to explain the diffusion of hybrid technology, therefore, it is important not only to
analyze farmers’ varietal adoption decisions, but also
to examine the factors that shape the incentives for
firms to produce and sell improved seed. The question of supply-side incentives is especially relevant in
developing countries, where private companies have
often been reluctant to enter into the production and
marketing of hybrid maize seed (Dowswell et al.,
1996).
269
The presence or absence of incentives to produce
and sell hybrid seed may be related to the stage of development of the local seed industry. Several authors
have advanced life cycle theories of seed industry
development in which national seed industries are
described as evolving in a path-dependent manner
through successive growth stages (Douglas, 1980;
Pray and Ramaswami, 1991; Rusike, 1995; Dowswell
et al., 1996; Morris and Smale, 1997; Morris et al.,
1998). According to the various life cycle theories,
the characteristics associated with the initial stages of
seed industry development mitigate against the diffusion of hybrid maize, because incentives to produce
and sell hybrid maize are not yet present. In the early
stages of seed industry development, maize producers
consist mainly of small-scale, subsistence-oriented
farmers who use mostly farm-saved seed retained
from their own harvest or obtained from neighbors.
Under these circumstances, there is no adequate market capable of sustaining firms looking to generate
profits through the production and sale of commercial seed. Not until the seed industry reaches more
advanced stages of development, when farmers understand the benefits of improved germplasm and
are willing to purchase seed on a regular basis, does
the effective demand for hybrid seed become strong
enough to support a commercial seed industry —
thereby paving the way for widespread diffusion of
hybrids. In short, the production and delivery of hybrid maize seed go hand-in-hand with the existence
of well-developed commercial seed industries.
In seeking to strengthen local production capacity
and improve the ability of public and private seed organizations to deliver improved seed to farmers, policy makers clearly have an interest in understanding
the factors that drive seed industries along the path
of development. To date, few efforts have been made
to test the validity of the life cycle theory of seed industry development. Similarly, little research has been
done to explore the relationship between the stage of
seed industry development and hybrid diffusion patterns. Empirical work in these areas is long overdue,
especially in LAC, where many governments have recently introduced policy reforms designed to improve
the performance of the seed sector. Without understanding the factors that influence seed industry development, policies already implemented and those to
be implemented in the future may prove ineffective.
270
J.L. Kosarek et al. / Agricultural Economics 26 (2001) 267–280
In this paper, we investigate the hybrid maize diffusion process. Using data from 18 countries in LAC,
we examine factors related to the successful diffusion
of hybrid maize. Following Griliches and Heisey et al.,
our model includes not only conventional demand-side
variables that are expected to influence farmers’ adoption decisions, but also supply-side variables that are
expected to affect the incentives for firms to produce
and sell hybrid seed. Since these variables were selected based on hypotheses generated by the life cycle
theory of seed industry development, our results can
be interpreted as a de facto test of the life cycle theory.
2. The life cycle theory of seed industry
development
A recurring theme in history is that science-based
industries often evolve along a growth path that can
be characterized as an industrial “life cycle” comprising successive stages of development (Rusike, 1995).
Several authors have used the life cycle framework to
characterize the development of seed industries (for
a review of selected studies, see Morris, 1998). Although the stages of development described by different authors differ in their details, each stage tends to
be associated with a unique characteristic combination of technology, organizational arrangements, and
institutional mechanisms. 4
Broadly speaking, four stages of development can
be distinguished within the life cycle of a national seed
industry:
1. Preindustrial stage: Maize is produced for subsistence purposes, i.e., it is cultivated to satisfy
household consumption requirements. Traditional
varieties (also known as landraces) are grown using farm-saved seed retained from the harvest or
procured through farmer-to-farmer exchange. In
the absence of formal research organizations, varietal improvement occurs as farmers replant seed
harvested from ears selected for their desirable
traits. Varietal information is obtained from other
farmers or based on direct observation of growing
plants or harvested ears.
4
Unlike other science-based industries, seed industries are unlikely to enter a phase of decline since they provide a basic necessity with few substitutes (Rusike, 1995).
2. Emergence stage: Maize continues to be cultivated mainly for subsistence purposes, although
surplus production may also be sold. Many farmers still grow traditional varieties, but some adopt
improved OPVs, and a few even experiment with
hybrids. 5 Public research organizations begin
conducting plant breeding research, and public
agencies begin producing commercial seed. Seed
production is carried out mainly on the farm,
although increasing numbers of farmers begin
purchasing commercial seed obtained from public
agencies. Public extension services play an active
role in educating farmers about the advantages of
improved germplasm and in disseminating varietal
information. Varietal improvement research and
seed production are performed mainly by public organizations, so the need for asserting and
enforcing intellectual property rights is negligible.
3. Expansion stage: Maize production becomes increasingly commercial, with more and more of the
crop being marketed. Most farmers have shifted to
hybrids, although a few continue to plant OPVs.
Private firms conduct varietal improvement research and distribute commercial seed in direct
competition with public agencies; very little seed is
still produced on the farm. In addition to producing
and selling seed, private companies provide technical information. Firms rely mainly on trade secrets
strategies to protect their intellectual property.
4. Maturity stage: Maize production is completely
commercial. Hybrids are the predominant technology, and most farmers regularly purchase commercial seed. Germplasm improvement research
continues to be conducted by public breeding
programs as well as by private seed companies,
but the two have become specialized, with the
public sector focusing on basic research and the
private sector concentrating on the development
5 Well-adapted hybrids can offer a significant performance advantage compared to OPVs due to heterosis (or hybrid vigor),
a phenomenon that results when genetically different parents are cross-pollinated. The effects of heterosis displayed in
first-generation progeny decrease in subsequent generations if seed
is saved and re-planted. Farmers who grow hybrids must therefore
purchase fresh seed each season to capture the benefits of heterosis. In contrast, farmers who grow OPVs can save seed from
their own harvest and replant it without experiencing a substantial
decrease in performance.
J.L. Kosarek et al. / Agricultural Economics 26 (2001) 267–280
271
A large body of empirical evidence suggests that
the diffusion of hybrid maize is jointly determined by
demand- and supply-side factors. Farm-level studies
make clear that the farmer’s decision to adopt hybrid
seed is affected by the expected profitability of the
technology, by risk considerations, and by the availability of seed, among other factors. At the same time,
industry-level studies suggest that at the aggregate
level, the diffusion of hybrid maize is affected by economic factors, by market structural conditions, and by
institutional considerations, all of which influence the
willingness of firms to supply seed.
Until now, few attempts have been made to relate
the empirical evidence on hybrid diffusion patterns to
a coherent conceptual framework. We believe the life
cycle theory of seed industry development provides
such a framework. Insights derived from the life cycle
theory can be used to formulate testable hypotheses
about the relationship between the diffusion of hybrid
maize and the presence or absence of supply and demand factors that temper the incentives to produce and
use hybrid seed. A number of these testable hypotheses are discussed in this section.
adopting hybrid maize seed. If the prices farmers must
pay for hybrid seed are high and/or if the prices farmers receive when they sell their grain are low, then
investment in hybrid seed may be unprofitable.
The adoption of hybrid maize is also affected by
risk factors. Two types of risk can affect the farmer’s
decision to adopt: subjective risk related to the uncertainty experienced by the farmer when confronted
with a new and unfamiliar technology, and objective
risk arising from uncontrollable environmental factors.
Farmers may be unwilling to invest in hybrid seed if
they are uncertain that the hybrid will outperform their
current variety. Even if they are convinced that the hybrid is capable of outperforming their current variety,
they still may not invest in hybrid seed if they fear
that the potential benefits might not be achieved due
to factors beyond their control, e.g., adverse weather.
A third demand-side factor affecting hybrid adoption rates relates to the availability of the technology.
Even if farmers know about the benefits of hybrids and
are willing to adopt, adoption will not take place if sufficient quantities of high-quality, affordable seed are
lacking. Alternatively, sufficient quantities of hybrid
seed may be available, but the type of germplasm may
not be appropriate. A recent study of maize varietal
use in Paraguay determined that the available hybrids
did not meet the consumption requirements of many
of the small-scale farmers; rather, the hybrids being
offered were more appropriate for the needs of commercial farmers (CIMMYT, 1992). Incompatibilities
between locally available cultivars and farmers’ consumption preferences have similarly limited the adoption of hybrid maize in Malawi (Smale et al., 1995).
3.1. Demand-side factors
3.2. Supply-side factors
The empirical literature on adoption makes clear
that farmers will not accept a new technology unless
they expect it to be profitable. Of the many factors
that affect the profitability of adopting hybrid maize,
one of the most important is the yield gain associated with the shift from the farmer’s current variety. If
the yield gain is expected to be modest, the additional
benefits (value of increased production) may not exceed the additional costs (higher seed costs, increased
management requirements), in which case farmers will
not adopt. Since benefits and costs depend on input
and output prices, these also affect the profitability of
Just as farmers will refrain from adopting hybrid
maize if the technology is seen to be unprofitable
or excessively risky, private firms will refrain from
supplying hybrid seed if they cannot recover their production and distribution costs and earn some profit.
The profitability of supplying hybrid maize seed is
of course affected by technical and economic factors
that determine seed production costs, but somewhat
less obviously it is also influenced by factors relating
to the structure of the seed market, the organization of
the seed industry, the cost of research innovation, and
the political importance of maize. Seed production
of locally-adapted cultivars. Seed production has
shifted to the private sector, and private companies
are the main source of technical information. Plant
varietal protection laws (PVPs) are in place, and legal systems function effectively at the international
level to allow protection of intellectual property.
3. Theoretical framework
272
J.L. Kosarek et al. / Agricultural Economics 26 (2001) 267–280
costs are directly influenced by technical factors that
impose physical limits on seed yields, irrespective
of the management ability of the grower. For example, seed yields of the highly inbred lines used to
produce single-cross hybrids tend to be low, making
single-cross hybrids relatively expensive to produce;
in contrast seed yields of the composite materials used
to produce top-cross hybrids and varietal hybrids tend
to be high, making these types of hybrids relatively
inexpensive to produce. Another technical factor that
can greatly affect seed production costs is the amount
of undersized or off-type seed which must be discarded; this is also highly variable and frequently
remains beyond the control of the seed grower.
Seed production costs are not determined exclusively by technical considerations; they are also influenced by economic factors, especially the prices of
key inputs such as labor, land, and capital. Field management operations (e.g., planting, detasseling, harvesting) tend to be labor intensive, increasing the cost
of seed production in countries where wage rates are
high. If seed production is contracted out to commercial seed growers, payments to growers must be high
enough to make seed production competitive with alternative uses of the land. In countries where access to
credit is limited, seed producers often must rely on expensive loans to finance variable and fixed costs. Prevailing factor prices thus exert an important influence
on the cost of seed production and affect the price at
which seed companies will sell their product.
The cost of seed production is an important determinant of profitability, but it is by no means the
only one. The structure of the seed market also has
important implications for potential seed suppliers.
Important aspects of market structure include not only
the size of the market, but also its composition. For
instance, if the market consists mainly of geographically concentrated, large-scale commercial farmers
who regularly purchase inputs in bulk, the greater
the profit potential. Conversely, if the market consists mainly of geographically dispersed, small-scale
subsistence farmers who do not purchase inputs on a
regular basis, the more modest the profit potential.
The incentives to supply hybrid maize seed can also
be affected by the organization and the composition
of the seed industry itself. Theoretically, a seed industry consisting of numerous small firms should be
more competitive than a market made up of a few large
firms. As the degree of competition increases, firms
that are battling to maintain or even expand market
share will feel pressure to develop innovative products
and to offer them at attractive prices. Although profit
margins for seed suppliers will be squeezed, farmers
will benefit by being offered a wider selection of technologies at competitive prices.
Regardless of the degree of seed industry concentration, the institutional arrangements that govern the
behavior of individual firms can also affect the supply
of hybrid maize. For instance, in the absence of PVPs,
firms may limit their investment in research and development, since they may not be able to appropriate
the benefits of research investments.
The incentives to supply hybrid seed can also be affected by the political importance of maize in a particular country. In countries in which the producer price
of maize is supported by the government for political reasons, farmers will be encouraged to produce
maize as a commercial crop, since they are assured an
adequate price at harvest. Alternatively, in countries
in which the price of maize is determined purely by
market forces, farmers may feel that commercial production of maize is too risky, and they will hesitate to
invest in improved technology.
Finally, the incentives to supply hybrid seed will be
influenced by the cost of research innovation. Maize
hybrids are bred to perform well in specific production environments. The cost of developing hybrids
adapted to a particular production environment may
be prohibitively high if the potential market is limited
and the likelihood of recouping the original research
investment is low. Thus, anything that will lower the
cost of research innovation is likely to improve the
incentives to supply hybrid seed. Since the cost of innovation will be lower in areas for which germplasm
has already been developed, firms will have greater
incentives to invest in hybrid research and development if well-adapted germplasm is readily and
inexpensively available.
4. Empirical framework
4.1. Estimation procedure
Following Heisey et al. (1998), and incorporating insights derived from the life cycle theory of
seed industry development, we analyze the diffusion
J.L. Kosarek et al. / Agricultural Economics 26 (2001) 267–280
of hybrid maize in LAC using cross-sectional data
from 18 countries. The dependent variable is the area
planted to hybrid maize, expressed as a percentage
of a country’s total maize area. This variable is not
continuous; rather, it is censored at the lower and upper limits (since it can take on values from 0 to 100).
Because the error term does not have zero mean, OLS
procedures may result in biased parameter estimates.
An appropriate approach for modeling censored dependent variables using maximum likelihood (ML) estimation procedures was developed by Tobin (1958).
The two-limit Tobit model is expressed as
y ∗ = β x + ε,
ε ∼ N (µ, σ 2 ),
(1)
where
y = L1
if y ∗ ≤ L1 ,
y = L2
if y ∗ ≥ L2 .
y = y∗
if L1 ≤ y ∗ ≤ L2 ,
In our model, y∗ represents the censored dependent
variable (in this case, percent maize area planted to
hybrids), x is a vector of independent explanatory variables representing characteristics of maize seed industries and of maize production in LAC, L1 = 0 is the
lower limit (no maize area planted to hybrids), L2 =
100 the upper limit (all maize area planted to hybrids),
and ε the error term with mean µ and constant variance σ 2 .
The likelihood function for the model is
L(β, σ |y, x, L1 , L2 )
L1 − β x 1 y − β x =
Φ
φ
σ
σ
σ
y=y ∗
y=L1
y=L2
L2 − β x
1−Φ
.
(2)
σ
Maximizing the logarithm of Eq. (2) provides the parameter estimates β̂ and σ̂ , which characterize the
variability of the dependent variable.
Denoting l1 = (L1 −β x/σ ) and l2 = (L2 −β x/σ ),
and writing the normal cumulative distribution functions as Φ1 = Φ(l1 ), Φ2 = Φ(l2 ) and the normal probability density functions as φ1 = φ(l1 ), φ2 = φ(l2 ),
the effects of changes in the explanatory variables (xj )
on the expected value of the dependent variable (y)
273
can be expressed as
−β̂
−β̂
∂E(y)
=0
φ1 − 100
φ2
∂xj
σ̂
σ̂
+(Φ2 − Φ1 )β̂
−β̂
−β̂
+β̂ xi
φ2 −
φ1
σ̂
σ̂
−β̂
−β̂
+σ̂ −l1
φ1 − (−l2 )
φ2
σ̂
σ̂
−β̂
=
[(0 − β̂ x) − σ̂ l1 ]φ1
σ̂
−β̂
[(1 − β̂ x) − σ̂ l2 ]φ2
−
σ̂
+(Φ2 − Φ1 )β̂ = (Φ2 − Φ1 )β̂
= (prob[0 < y ∗ < 100])β̂.
(3)
Eq. (3) provides the marginal effects of changes in
the seed industry characteristics and maize sector explanatory variables on the proportion of maize area
planted to hybrids.
4.2. Variables
Seven explanatory variables are used to represent
economic, institutional, and policy factors that are expected to influence the diffusion of hybrid maize (see
Table 1). Some of these variables were chosen on the
basis of conventional profitability-based explanations
of farmer adoption behavior, while others were included to test the importance of supply-side factors as
suggested by the life cycle theory.
PRICE: Seed price (expressed as the seed-to-grain
price ratio to facilitate inter-country comparisons) is
expected to affect the profitability of adopting hybrid
maize. As with any purchased input, the higher the
price, the lower the expected level of use. The sign on
this variable is expected to be negative.
PRICE2: Seed price squared is included to allow
the relationship between seed prices and hybrid adoption levels to vary according to the stage of seed
industry development. Although economic theory
suggests that higher seed prices can be expected to reduce demand for hybrids, casual observation suggests
that seed prices tend to rise through time as the seed
274
J.L. Kosarek et al. / Agricultural Economics 26 (2001) 267–280
Table 1
Identification of variables used in model
Variable description
Variable name
Measurement
Dependent variable
Adoption of hybrid maize seed
AREA HYB
Area planted to hybrids as a
proportion of national maize area
Independent variables
Maize seed price (2)
(1) PRICE
Commercialization of the maize sector
(2) PRICE2
MARKET
Legislation protecting intellectual property
Government support policies
Level of private-sector seed industry activity
PVP
NPC
PRIVATE
Cost of research innovation
GERMPLASM
Seed-to-grain price ratio of the
most commonly sold type of seed
PRICE variable squared
Percent of total maize production
that is marketed
PVPs established? (1: yes; 0: no)
NPC for maize (%)
Percent of total industry seed sales
made by the private sector
Percent of all seed sold containing
CIMMYT germplasm
industry develops and as the area planted to hybrids
increases. This pattern may be explained by arguing
that initially low seed prices may be needed to encourage farmers to try hybrids, but once farmers have
come to appreciate the benefits of the technology,
they are willing to pay higher prices. Price increases
occurring at different stages in industry development
may also reflect costs associated with improvements
in the quality of the hybrids being sold. The sign on
this variable is expected to be positive.
NPC: Nominal protection coefficients (NPCs) for
maize are included to provide a rough measure of the
level of policy protection afforded to the maize sector in each country. Although other measures of protection are more comprehensive — e.g., the effective
protection coefficient (EPC) and the producer subsidy
equivalent (PSE) — the data needed to calculate these
other measures were unavailable or incomplete. Since
increased government protection for the maize sector
enhances the expected profitability of adopting hybrid
maize seed, the sign on this variable is expected to be
positive.
MARKET: The percent of maize production that
is marketed is included as another measure of seed
industry profitability. As agriculture becomes more
commercialized, farmers increasingly base input use
decisions on profitability considerations, and purchased inputs rapidly replace those once non-traded
(Pingali, 1997). As the volume of purchased seed in-
Expected sign
−
+
+
+
+
+
+
creases, seed suppliers find it worthwhile to develop
and promote a greater range of improved technologies. The degree of commercialization of the maize
sector is therefore expected to be positively related to
the diffusion of hybrids. The sign on this variable is
expected to be positive.
PVP: Intellectual property rights are important to
the development of the seed industry because they
can strengthen incentives for firms to invest in plant
breeding research. Although hybrid maize technology
traditionally has been protected through trade secrets
strategies, advances in technology have made it easier
and cheaper for competitors to identify and replicate
successful commercial hybrids. Without effective intellectual property laws, it is becoming increasingly
difficult to prevent free riders from appropriating benefits from research investments made by others. A
dummy variable is used to indicate the presence or absence of PVPs in each country. Since the presence of
intellectual property laws presumably encourages increased investment in hybrid seed research, the sign
on this variable is expected to be positive.
PRIVATE: According to the life cycle theory, as the
seed industry matures, the private sector plays an increasingly prominent role in varietal research and seed
production. As private firms become more prevalent,
they compete for potential customers by stepping up
their marketing efforts. Their efforts to develop product and brand awareness, trust, and customer loyalty
J.L. Kosarek et al. / Agricultural Economics 26 (2001) 267–280
stimulate further adoption of hybrid maize technologies. This effect is captured by means of a variable
that expresses private-sector seed sales as a percent of
total industry seed sales. The sign on this variable is
expected to be positive.
GERMPLASM: The availability of improved
germplasm developed by public breeding programs
helps to stimulate private-sector investment in plant
breeding, because it is considerably less costly for
private seed companies to develop new hybrids when
public germplasm is freely available (López-Pereira
and Garcia, 1997). The International Maize and
Wheat Improvement Center (CIMMYT) is the leading
public maize breeding organization in the developing
world. Since CIMMYT germplasm is freely available
to private seed companies in LAC, one measure of
the availability of publicly developed germplasm is
the percent of commercial seed sold that contains
CIMMYT germplasm. By making use of CIMMYT
germplasm, private firms can reduce their research
innovation costs, which presumably encourages them
to invest more in hybrid research, leading to an expanded range of products. The sign on this variable is
expected to be positive.
4.3. Data
As part of an ongoing effort to document the impacts of international maize breeding research, CIMMYT periodically surveys organizations engaged in
maize research and maize seed production activities.
In 1996–1997, staff of the CIMMYT Economics Program carried out a comprehensive survey of public
and private maize seed organizations located throughout LAC. Data were collected from 208 organizations
in 18 countries, including almost all of the organizations in those countries with maize breeding programs.
Collectively, these organizations produced over 97%
of the commercial maize seed sold in LAC in 1996
(Table 2). 6
Data collected through the survey, as well as data
downloaded from the FAOSTAT online database, were
used to estimate the variables included in the model.
Descriptive statistics for the variables are presented in
Table 3.
6 Additional information about the survey appears in Morris and
López-Pereira (1991).
275
5. Results
The tobit maximum likelihood estimates (MLE),
and their corresponding marginal effects (Eq. (3)) are
presented in Table 4 (columns 1 and 3). The marginal
effects measure the effects of a one-unit change in
the explanatory variable on the percent of national
maize area planted to hybrids, given the censoring of
the dependent variable. For comparative purposes, ordinary least-squares (OLS) results are also presented
(column 2).
Preliminary examination of the OLS residuals of
the estimated relationship suggested the possibility of
heteroskedasticity. In OLS models, heteroskedasticity
results in unbiased but inefficient parameter estimates.
In tobit models, parameter estimates will be biased
and inefficient. To assess possible heteroskedasticity
problems, the tobit model was re-estimated under
the assumption that σi2 = exp(α xi ), where the appropriate xi ’s (PVP and GERMPLASM) were identified from examination of the OLS residuals. The
parameter estimates did not change qualitatively,
and the log-likelihood ratio statistic (−2(−60.297 −
(−58.011)) = 4) used to assess the significance
of the heteroskedastic terms was less than the critical value (χ 2 statistic with 2 degrees of freedom,
5.99), supporting the validity of the results from the
straightforward tobit procedure presented in Table 4.
Several points emerge from the results. First, the
tobit and OLS estimated coefficients are very similar. Likewise, the tobit estimated coefficients and their
marginal effects are virtually the same. Both of these
results appear to stem from the apparent limited degree of censoring of the dependent variable (only two
observations at the limit). Second, the significance of
the estimated relationships is high. The overall significance of the tobit equation is measured by the Wald
statistic, which follows a χ 2 distribution. Based on the
adjusted R2 and the Wald test (with 9 degrees of freedom and a critical value of 16.92), the overall significance of the relationships is acceptable. 7 All of the
7
To explore the robustness of the findings, regressions also were
run on subsets of the data from which the cases corresponding
to the extreme values of the dependent variable were omitted.
The estimated coefficients for all explanatory variables remained
significant at the 10% level and retained the signs except for the
coefficient on PRIVATE (which was no longer significant at the
10% level), suggesting that the model is reasonably robust.
276
J.L. Kosarek et al. / Agricultural Economics 26 (2001) 267–280
Table 2
Coverage of the 1997 Latin America maize impacts surveya
Public organizations
surveyed
Private companies
surveyed
1996 maize seed
sales (000 t)
Percent of formal
maize seed market
Central America
Costa Rica
El Salvador
Guatemala
Honduras
Nicaragua
Panama
7
1
1
1
2
1
1
39
4
6
5
13
9
2
5.5
0.1
2.1
1.5
1.0
0.5
0.3
90
100
91
86
95
95
90
Caribbean
Cuba
Dominican Republic
Haiti
3
1
1
1
4
0
2
2
2.4
1.6
0.5
0.3
95
100
100
65
Mexico
11
18
32.6
97
Central America, Caribbean, Mexico
21
61
40.5
96
Andean Zone
Bolivia
Colombia
Ecuador
Peru
Venezuela
11
5
1
2
2
1
63
15
9
7
15
17
21.1
2.6
3.2
2.8
1.6
10.9
96
88
95
96
88
100
Southern Cone
Argentina
Brazil
Chile
Paraguay
Uruguay
4
1
1
–
2
–
48
14
24
–
10
–
219.4
64.9
151.8
–
2.7
–
98
94
99
–
93
–
South America
15
111
240.5
98
Latin America
36
172
281.0
97
a
Source: Morris and López-Pereira (1991).
Table 3
Descriptive statistics
Variable
Range
Minimum
Maximum
Mean
Standard deviation
Variance
AREA HYP
PRICE
PRICE2
MARKET
PVP
NPC
PRIVATE
GERMPLASM
98.99
38.37
1599.91
54.31
1.00
190.12
100.00
92.11
0.00
1.67
2.78
45.58
0.00
40.60
0.00
7.89
98.99
40.03
1602.69
99.89
1.00
230.73
100.00
100.00
32.28
12.32
245.66
78.45
0.33
106.70
74.03
79.20
28.03
9.98
389.73
15.75
0.49
50.06
30.98
30.89
785.72
99.50
151891.00
248.20
0.24
2505.56
959.83
953.95
J.L. Kosarek et al. / Agricultural Economics 26 (2001) 267–280
277
Table 4
Tobit results, OLS comparisons, and marginal effects
Tobit
OLS
Constant
−340.040
PRICE
−7.530 (−5.992)
PRICE2
0.218 (5.710)
NPC
0.895 (5.648)
MARKET
2.293 (8.205)
PVP
33.150 (3.548)
PRIVATE
0.290 (1.522)b
GERMPLASM
1.291 (3.795)
Wald test: χ 2 (tobit) 112.51, d.f. = 9, adjusted R2 (OLS) 0.801,
σ̂ = 10.04, proportion of the sample between the limits = 88.9
−316.870
−7.074
0.201
0.864
2.218
30.670
0.319
1.095
(−7.563)a
a
b
Marginal effect
(−6.090)
(−4.684)
(4.482)
(4.492)
(6.549)
(2.739)
(1.359)b
(2.863)
−339.682
−7.521
0.218
0.894
2.290
33.115
0.289
1.290
The numbers in parentheses are t-values.
Significant at the 0.10 level. All others, significant at the 0.01 level.
estimated coefficients are significant at the 1% level
(using a one-tailed test), with the exception of PRIVATE, which is significant at the 10% level.
In the tobit model, the signs on all of the estimated coefficients are consistent with the hypotheses
regarding effects of the demand-side and supply-side
variables on the diffusion of hybrid maize. On the
demand-side, high seed-to-grain price ratios (PRICE)
have a large and significant negative effect on the area
planted to hybrid maize. The relationship is not expected to be linear, however, so a second variable,
PRICE2, was included to capture non-linearities in
the relationship. Combining the estimated coefficients
of the two price variables, a one-unit increase in the
seed-to-grain price ratio at the mean of PRICE is associated with a 6.6% decrease in the area planted to
hybrid maize. These results indicate that seed price
is an important determining factor of hybrid diffusion
and that lower seed-to-grain price ratios in the initial stages of development are positively related to the
spread of hybrid maize.
The strength of government support for maize production, measured by the NPC for maize, positively
influences the spread of hybrid maize. Every one-unit
increase in the NPC for maize (in this case, one unit is
equal to 1%, since the NPC ratios were converted into
percentages) is associated with an increase of 0.89%
in the area planted to hybrid maize.
On the supply-side, higher seed-to-grain prices
(PRICE2) are associated with greater levels of hybrid
use during the later stages of the seed industry development, indicating that as farmers become more
familiar with hybrids and learn to appreciate their
benefits, they are willing to pay higher prices. Also,
the fact that seed prices are higher at increased levels
of hybrid use indicates that as demand strengthens,
seed companies can sell more sophisticated, higher
quality hybrids.
Increased commercialization, measured by the percent of total maize production that is marketed (MARKET), is also associated with greater use of hybrids,
suggesting that seed companies respond to opportunities for achieving greater sales volumes by targeting
areas with large market potential. The marginal effect
of this variable indicates that a 1% increase in the proportion of total maize production that is marketed is
associated with a 2.29% increase in the area planted
to hybrid maize.
The percent of total seed sales in a country coming
from the private sector (PRIVATE) is positively related
to the spread of hybrid maize, suggesting that as private sector participation in the seed industry increases,
enhanced competition leads to greater marketing and
promotion of hybrids. The marginal effect of this variable indicates that a 1% increase in the proportion of
total seed sales attributable to the private sector is associated with a 0.29% increase in the area planted to
hybrid maize.
Greater use of CIMMYT germplasm (GERMPLASM) is positively related to the spread of hybrid
maize, indicating that the availability of germplasm
from public breeding programs fosters innovation
by reducing research costs for private firms. A 1%
increase in the percentage of seed sold containing
278
J.L. Kosarek et al. / Agricultural Economics 26 (2001) 267–280
CIMMYT germplasm is associated with a 1.29%
increase in the area planted to hybrids.
The presence of PVPs is positively related to the
area planted to hybrid maize, suggesting that the presence of government legislation that protects firms’
intellectual property strengthens incentives to supply
hybrid technology.
In the initial stages of the analysis, other explanatory variables were included in the model. Following Heisey et al. and Griliches, a farm size variable
was included to reflect the fact that sales costs decline for suppliers of hybrid seed as the volume of
each seed transaction increases. No significant relationship was found between this variable and hybrid
maize diffusion, however. An infrastructure variable
(percentage of roads in each country that are paved)
also was included in an effort to capture differences
between countries in the cost of distributing agricultural inputs, but once, again no significant relationship
was found between this variable and the diffusion of
hybrid maize.
6. Conclusions and implications
Our results indicate that at the aggregate (country)
level, the diffusion of hybrid maize in LAC can be
explained in terms of a combination of economic,
institutional, and policy factors. Consistent with earlier findings reported by Griliches and Heisey et al.,
we conclude that diffusion is influenced not only
by demand-side factors that affect the incentives for
farmers to adopt improved germplasm, but also by
supply-side factors that affect the incentives for firms
to produce and sell hybrid seed. In addition to validating conventional profitability-based explanations
of farmer adoption behavior, our findings provide
strong empirical support for the life cycle theory of
seed industry development and confirm its usefulness
in explaining the spread of hybrid maize in LAC.
In terms of drawing policy implications, these
results must be interpreted with care. The life cycle theory posits that seed industry development is
influenced by a large number of interrelated and
time-dependent factors, so not only the nature but
especially the sequencing of policy interventions is
likely to be important. One thing seems clear, however: if the technology diffusion process is to be
accelerated, policy makers need to ensure an environment in which it is profitable not only for farmers
to adopt improved germplasm, but also for the seed
industry to produce and sell high-quality seed.
What policy options are available to achieve these
two objectives? On the demand-side, policy interventions to promote increased use of hybrid seed are
relatively limited, since the profitability of adopting
hybrids depends to a considerable extent on factors
that are difficult for policy makers to control (e.g.,
agro-climatic conditions, factor endowments, proximity to international markets). In some countries, governments have attempted to stimulate technical change
by increasing the profitability of maize production, either directly (by supporting producer prices) or indirectly (by restricting grain imports through quotas or
tariffs), but in the long run such protectionist policies
often turn out to be unsustainable because they impose
an unacceptable fiscal burden. Whether or not policies
that are geared toward supporting producers are sustainable in the long run may depend on the specific
characteristics of a country’s economy, as well as the
nature and the extent of the support. Experience suggests that a more effective way to increase demand
for hybrids is to invest in programs designed to educate farmers about the potential benefits of adopting
improved seed. The role of the state in providing technical information is particularly important during the
early stages of seed industry development, when the
market for commercial seed is still poorly developed
and private firms have incentives to invest in extension
activities.
On the supply-side, policy measures needed to accelerate the diffusion of hybrid technology are more
obvious. In order to foster the emergence of a flourishing seed industry, policy makers should consider
steps to encourage the development and growth of private firms, provide incentives for the continuing role
of the public sector in research activities, improve the
efficiency of the grain and seed marketing systems,
and promote commercialization of the maize sector in
general. In pursuing these objectives, timing is likely
to be important. For example, the existence of an efficient grain marketing system in and of itself is unlikely
to stimulate widespread marketing of hybrid seed if
an unfavorable business climate continues to stifle the
activities of private companies. Experience suggests
that large firms operating in a commercial marketing
J.L. Kosarek et al. / Agricultural Economics 26 (2001) 267–280
environment can more easily take advantage of potential economies of scale in varietal improvement research, so institutional arrangements that encourage
specialization and permit a more efficient transfer of
improved germplasm from public breeding institutions
to the private sector are also likely to be important.
Throughout LAC, national maize seed industries
are being transformed as private firms increasingly
take over activities once dominated by public agencies. Privatization is bringing many welcome changes,
but it is important not to lose sight of the fact that
successful privatization cannot take place in a vacuum. Our findings suggest that the public sector still
has a vital role to play in creating the proper enabling
environment needed by an increasingly privatized
seed industry to operate efficiently. Policy initiatives
are urgently needed in many countries to facilitate the
transition to more commercial hybrid seed marketing
systems. Further, continued support for basic research
by public breeding institutions seems entirely consistent with the objective of stimulating the spread
of productivity-enhancing hybrid technologies. And
while the private sector can be expected to attend to the
technology requirements of commercial farmers, public research will be needed to ensure that the needs of
small-scale, subsistence-oriented farmers also are met,
especially in difficult production environments and in
areas in which the welfare of the population is highly
dependent on maize production. Only by striking an
appropriate balance between public- and private-sector
participation will it be possible to propel national
seed industries through successive stages of development, stimulate increased productivity throughout
the region’s maize sectors, and help farmers keep
pace with projected strong growth in demand for
maize.
References
Byerlee, D., Morris, M.L., López-Pereira, M.A., 1993. Hybrid
maize and the small-scale farmer: economic and policy issues
for Asia. Paper Presented at the Fifth Asian Regional Maize
Workshop, November 15–20, 1993, Hanoi and Ho Chi Minh,
Socialist Republic of Vietnam.
CIMMYT, 1992. 1991–1992 CIMMYT world maize facts and
trends: maize research investment and impacts in developing
countries. International Maize and Wheat Improvement Center
(CIMMYT), Mexico City, Mexico.
279
Douglas, J.E. (Ed.), 1980. Successful Seed Programs: A Planning
and Management Guide. Westview Press, Boulder, CO.
Dowswell, C.R., Paliwal, R.L., Cantrell, R.P., 1996. Maize in the
Third World. Westview Press, Boulder, CO.
FAO, 2000. FAOSTAT Online Database. Food and Agriculture
Organization of the United Nations, Rome, Italy.
Feder, G.R., Just, R.E., Zilberman, D., 1985. Adoption of
agricultural innovations in developing countries: a survey. Econ.
Dev. Cult. Change 33, 255–298.
Gerhart, J., 1975. The diffusion of hybrid maize in Western
Kenya. International Maize and Wheat Improvement Center
(CIMMYT), Mexico City, Mexico.
Griliches, Z., 1957. Hybrid corn: an exploration in the economics
of technological change. Econometrica 25, 501–522.
Heisey, P.W., Morris, M.L., Byerlee, D., López-Pereira, M.A.,
1998. Economics of hybrid maize adoption. In: Morris, M.L.
(Ed.), Maize Seed Industries in Developing Countries. Lynn
Rienner, Boulder, CO, pp. 143–158.
Kumar, S.K., 1994. Adoption of hybrid maize in Zambia: effects
on gender roles, food consumption, and nutrition. Research
Report No. 100. International Food Policy Research Institute,
Washington, DC.
López-Pereira, M.A., Garcia, J.C., 1997. The maize seed
industries of Brazil and Mexico: past performances and
future prospects. CIMMYT Economics Program Working Paper
97-02. International Maize and Wheat Improvement Center
(CIMMYT), Mexico City, Mexico.
Morris, M.L. (Ed.), 1998. Maize Seed Industries in Developing
Countries. Lynne Rienner, Boulder, CO.
Morris, M.L., López-Pereira, M.A., 1991. Impacts of maize
breeding research in Latin America, 1966–1997. International
Maize and Wheat Improvement Center (CIMMYT), Mexico
City, Mexico.
Morris, M.L., Smale, M., 1997. Organization and performance
of national maize seed industries: a new institutionalist
perspective. CIMMYT Economics Program Working Paper
97/05. International Maize and Wheat Improvement Center
(CIMMYT), Mexico City, Mexico.
Morris, M.L., Smale, M., Rusike, J., 1998. Maize seed industries:
a conceptual framework. In: Morris, M.L. (Ed.), Maize Seed
Industries in Developing Countries. Lynn Rienner, Boulder, CO,
pp. 35–54.
Pingali, P., 1997. From subsistence to commercial production
systems: the transformation of Asian agriculture. Am. J. Agric.
Econ. 79, 628–634.
Pray, C.E., Ramaswami, B., 1991. A Framework for Seed Policy
Analysis in Developing Countries. International Food Policy
Research Institute (IFPRI), Washington, DC.
Rauniyar, G.P., Goode, F.M., 1992. Technology adoption on small
farms. World Dev. 20 (2), 275–282.
Rosegrant, M.W., Agcaoli-Sombilla, M., Perez, N.D., 1995.
Global food projections to 2020: implications for investment.
Food, Agriculture, and the Environment Discussion Paper
5. International Food Policy Research Institute (IFPRI),
Washington, DC.
Rusike, J., 1995. An institutional analysis of the maize seed
industry in southern Africa. Ph.D. Thesis. Department of
280
J.L. Kosarek et al. / Agricultural Economics 26 (2001) 267–280
Agricultural Economics, Michigan State University, East
Lansing, MI.
Smale, M., Kaunda, Z.H.W., Makina, H.L., Mkandawire,
M.M.M.K., Msowoya, M.N.S., Mwale, D.J.E.K., Heisey,
P.W., 1991. Chimanga Cha Makolo, hybrids, and composites:
an analysis of farmers’ adoption of maize technology in
Malawi, 1989–1991. CIMMYT Economics Working Paper
91/04. International Maize and Wheat Improvement Center
(CIMMYT), Mexico City, Mexico.
Smale, M., Heisey, P.W., Leathers, H.D., 1995. Maize of
the ancestors and modern varieties: the microeconomics of
high-yielding varietal adoption in Malawi. Econ. Dev. Cult.
Change 43 (2), 351–368.
Tobin, J., 1958. Estimation of relationships for limited dependent
variables. Econometrica 26, 24–36.
Walker, T.S., 1981. Risk and Adoption of Hybrid Maize in El
Salvador, Vol. 18, No. 1. Food Research Institute Studies,
Stanford University, Stanford, CA, pp. 59–83.