AAMP Training Materials
Module 2.1: Factors Affecting Smallholder
Commercialization
T.S. Jayne & William Burke, (MSU)
jayne@msu.edu, burkewi2@msu.edu
Contents
• Agricultural commercialization
• Definition
• Indicators – Labor productivity
• Pros and cons
• Agricultural commercialization and the Structural
Transformation framework
• Constraints and challenges facing the promotion of
agricultural commercialization
• Market participation modeling: strengths, weaknesses,
and an example of an applied model
Agricultural Commercialization (Definition)
• The degree to which agricultural production and
marketing is based on market exchange and
specialization vs. internalized within the household
Agricultural Commercialization (Indicators)
• Household Commercialization Index (HCI)
– HCI = value of farm sales / value of farm production
– Ranges from 0 – 1.0
– The higher the number, the more commercialized the household
• Existence of / degree of reliance on:
– Factor markets (fertilizer, labor, etc)
– Financial markets (loans)
– Output markets (crop sales)
• Rising labor productivity in agriculture…
Labor Productivity
Where:
Y
L
=
Y*A
A L
Y/L = Labor productivity
A/L = Land to Labor Ratio
Y/A = Yield
Y = Net output
L = Labor
A = Area
Commercialization: Some Pros and Cons
• Pros:
‒ Can achieve growth in net output per unit area (Y/A) with shifts
in cropping patterns toward higher value crops
‒ Can be the result of input intensification, as long as the value of
output exceeds the costs of the additional inputs.
• Cons:
‒ Dependence on markets, both for the commercialized crops as
well as the food market.
‒ Competition from large-scale farming
• Important: Commercialization only works if the food
market is functioning well!
Agricultural commercialization and the
Structural Transformation framework
• Structural transformation / Demographic Transition
– Based on economic history of Asia and Europe (B. Johnston, J.
Mellor, Binswanger, Lipton)
– In early stages of development, 80% of population is primarily
engaged in staple food production
– Broad-based rural productivity growth triggers demographic
transition
Ag Commercialization & Structural
Transformation
• There is a symbiotic relationship between rural farms
and towns
– Urban areas provide a market for surplus farm output
– Farmers with cash generate demand for urban employment
– As demand for off-farm jobs rises people migrate to towns
– Increased urbanization “pulls” rural labor to urban areas
– birth rates decline and levels of education rise
FUNDAMENTALLY A SMALLHOLDER-LED MODEL!
Ag Commercialization & Structural
Transformation
1
2
3
4
Key elements of structural transformation
• Growth processes that effectively reach a large
proportion of the population
• Especially the poor – equitable growth
• That being said, broad based equitable growth is difficult
to achieve.
Extreme concentration of marketed maize
output – Malawi, 2008/09
Source: Agricultural Inputs Support Survey (n=1904 farm households), sample frame
from National Statistical Office, Government of Malawi
Challenges to commercialization
• The major challenge to the success of smallholder-led
commercialization strategies is to address the asset
constraints that prevent such a large % of the rural
population from being able to respond to growth
opportunities and incentives.
hectares
Most smallholders lack the land and other
resources to produce a surplus
Relationship between landholding size and
farm household income
ETHIOPIA
Log(Per Capita Income)
6.2
KENYA
9.8
RWANDA
4.4
4.2
5.8
9.4
4.0
5.4
3.8
9.0
0
.25
.5
Ha
.75
1
MOZAMBIQUE
4.0
0
.25
.5
Ha
.75
1
ZAMBIA
3.8
3.6
3.5
3.4
3.0
3.2
0
.25
.5 .75
Ha
1
0
.25 .5 .75
Ha
1
Per Capita Land Access (Ha)
Note: The vertical lines are drawn at 25th, 50th, and 75th percentiles of per capita land owned for each
country. The top 5 percent of observations are excluded from the graphs because lines are sensitive to a
few extreme cases.
0
.25
.5
Ha
.75
1
Labor Productivity
Where:
Y
L
=
Y*A
A L
Y = Output
L = Labor
A = Area
Y/L = Labor productivity
A/L = Land to Labor Ratio
Y/A = Yield
Evidence shows that broad-based ag
commercialization is associated with:
• Relatively equitable initial distribution of productive
assets and resources within society
• Agricultural growth in agrarian-based economies
• Public support for investments that most of the
population can take advantage of: infrastructure, seed
research, extension programs
Evidence shows that inequitable
commercialization is associates with:
• Highly concentrated initial distribution of productive
assets and resources within society
– e.g., latifundia-type landholding systems
• Elite capture of political process
• Use of public funds to invest in ways that are primarily
appropriated by elites
– Marketing board operations that raise prices with regressive
income distributional effects
– Input subsidy programs that are disproportionately targeted to
better-off farmers (not in all cases though)
Now let’s move to methods for assessing
the factors associated with smallholders’
ability to participate in markets
(commercialization)
Econometric Modeling Example
Example of a basic econometric model
y    1 x1   2 x2  
Talktime   1 years   2 contacts 
• Assumes the “dependent” variable y, is continuous (e.g.
every mobile phone owner uses at least some talktime.
Market participation models
• Modeling market participation (e.g. how much maize is
bought/sold per household in Zambia), can be a bit tricky
because not all households sell their crops.
• Often, “Two-stage” or “Double Hurdle” models are used
to address this issue.
• These models include two steps reflecting the dual
decision making process:
Decision 1: Whether to participate
Decision 2: How much to buy or sell
2-Tiered Market Participation Model (Ex 1)
Producing
Household
Decision 1
Participate in
the Market?
Do not Buy
or Sell
Market
Participant
Decision 2
How much will
market
participants buy
or sell
0
Net Purchases
Net Sales
2-Tiered Market Participation Model (Ex. 2)
Producing
Household
Decision 1
Participate in
the market as a
buyer, seller or
not at all?
Net Buyer
Decision 2
If I participate,
how much am I
buying and
selling?
0
Do not Buy
or Sell
Net Seller
0
Net Purchases
Net Sales
Market participation models
• The previous models include two steps reflecting the
dual decision making process:
Decision 1: Whether to participate
Decision 2: How much to buy or sell
• The weakness of the two-tiered models discussed so far
is that it requires all observations to be producers
– Can’t extrapolate to entire country
– May under-estimate the effects of a given policy on marketing
behaviour (i.e. if policy affects the decision to produce)
• The Triple Hurdle Model (Burke, 2009) can give a more
comprehensive perspective
Triple Hurdle Model: Market Participation
with Production Decision
Nationally Representative Sample
Decision 1
Non-Producers
Whether to
produce
Producing
Household
Decision 2
Participate as
buyer, seller or
not at all
Autarkic
Net Buyers
Net Sellers
Quantity
Bought
Quantity
Sold
Decision 3
Quantity bought
or sold among
participants
Insights from the Triple Hurdle Market
Participation Model
• The Triple-Hurdle Model provides the probability of:
–
–
–
–
Not producing
Producing, but also being a net buyer.
Producing, but not participating in the market
Producing and selling
• Expected value of market participants net purchases or
net sales
• Expected value of sales and purchases of any
household (per household expected values)
Kenyan dairy example
• About 85% of East Africa’s 3.5 million dairy cattle in are
raised in Kenya
• Compared to Maize
–
–
–
–
Although nearly all households produce maize,
Less than ½ of maize producers sell to generate cash
On the contrary, about 70% of dairy producers are net sellers
Top 10% of households generate ~ 80% of maize revenue,
versus ~ 55% of dairy revenue
Kenyan dairy example (2)
• Since private dairy purchasing enterprises became legal,
there are four principle buyers of dairy:
–
–
–
–
Kenya Creamery Company
Other co-ops
Private processors
Informal dairy “hawkers” tolerated, not legal
• Despite growth, domestic supply is outpaced by
domestic demand and export.
• Why?
What explains market participation?
•
•
•
•
•
Transfer cost determinants
Endowments
Access to production inputs
Market Prices
Production shocks
Triple Hurdle Model Simulation (Exercises)
• Open the Excel workbook corresponding to this
presentation
• Read the information in the [NOTES] sheet carefully
• You will use the output from a triple-hurdle model of
market participation in the Kenyan Dairy sector to
evaluate the impacts of various shocks on dairy market
participation
• Changes to yellow-highlighted values in [Simulation
values] sheet create changes in the probabilities listed in
row 3 of the [Simulation results] sheet
Triple Hurdle Model Simulation
(Exercise 1: Baseline)
a) At the mean data values provided, what is the
probability that the average household does not
produce dairy?
b) What is the probability that they are a producer, but net
buyer of dairy?
c) What is the probability that they produce, but neither
buy nor sell dairy?
d) What is the probability that they are a producer and net
seller of dairy?
e) Finally, what is the unconditional expected value of net
sales at the data means?
Triple Hurdle Model Simulation
Exercise 2: Distance to Electricity
a) Holding all other factors at their data means, what is the
expected value of net sales for a given household that
is 10 kilometers from the nearest source of electricity?
(Change cell 22E to 10 in [Simulation values] sheet)
b) What is the expected value of net sales for a given
household that has electricity nearby? (Change cell 22E
to 0)
c) Why would a dairy farmer closer to electricity be
expected to sell more dairy?
Triple Hurdle Model Simulation
Exercise 3: Land Holdings
Return cell E22 to it’s data mean value (4.17).
a) What is the likelihood of being a producer when lagged
acres owned is 15? (Cell E30 = 15)
b) What is the likelihood of being a producer when lagged
acres owned is 0.5? (Cell E30 = 0.5)
c) What does this tell us about dairy farming in Kenya, and
what are the policy implications of this result?
Triple Hurdle Model Simulation
Exercise 3: Land Holdings
Reset lagged acres owned to the mean (Cell E30 = 3.5).
a) Simulate an observation in which no organized
purchasers are present. (Change E25:E28 = 0) What
are the unconditional expected net sales?
b) Is it feasible that this household could be a net seller?
How?
c) The presence of which buying agent is correlated with
the highest unconditional expected net sales, and how
many litres of milk per year are sold by the average
household?
d) What are the policy implications of this finding?
Resources
• Bellemare, M.F., and C.B. Barrett. 2006. “An Ordered Tobit Model of
Market Participation: Evidence from Kenya and Ethiopia.” American
Journal of Agricultural Economics 88: 324-337
• Burke, W.J. 2009. "Triple Hurdle Model of Smallholder Production and
Particiaption in Kenya's Dairy Sector." MS Thesis. Department of
Agriculture Food and Resource Economics. Michigan State University.
• Goetz, S.J. 1992. “A Selectivity Model of Household Food Marketing
Behavior in Sub-Saharan Africa.” American Journal of Agricultural
Economics 74: 444-452.
• Key, N., E. Sadoulet, and A. DeJanvry. 2000. “Transactions Costs and
Agricultural Household Supply Response.” American Journal of Agricultural
Economics 82:245-259.