Livestock Research for Rural Development 28 (4) 2016 Guide for preparation of papers LRRD Newsletter

Citation of this paper

Factors influencing livestock marketing dynamics in Zambia

Mary Lubungu

University of Hohenheim, Institute 490C, 70599 Stuttgart, Germany
marylubungu@gmail.com

Abstract

The livestock sector has significant opportunities for economic growth and poverty reduction, especially among the rural farmers in developing countries. However, smallholder livestock producers are characterized by low levels of market participation. Using a three year panel data of 524 cattle owning households drawn from a national representative survey, this paper estimates a multinomial logit to identify the factors influencing the moving into and out of livestock market as well as being a consistent cattle seller.

Results indicate that education of household head, ownership of assets, market access and institutional factors as well as household shocks influence the likelihood of a household moving into and out of cattle markets. In addition the level of income source diversification influences the movement into and out of cattle markets.

Key words: cattle, market participation, panel data, multinomial logit


Introduction

Rapid urbanization, human population growth, coupled with sustained income growth and the emergence of an urban middleclass income group are transforming the agri-food industry in the developing countries. The most remarkable effect of this transformation is a sustained increase of consumption of animal proteins by urban population (Delgado 2003; Delgado 1999; Reardon et al 2014). The level of growth in meat consumption, however, varies across different continents (Livestock data innovation in Africa 2014). In Africa, for instance, estimates of meat consumption are expected to increase from 10.5 million tons by 2007 to 34.8 million tons by 2050 with the annual growth rate of about 2.8 percent (Livestock data innovation in Africa 2014). And Delgado (1999) projects that about 60% of all meat supplied will come from the smallholder and backyard producers. Therefore, the livestock sector has the potential to serve as a poverty reducing vehicle for high persistent rural poverty especially in the developing countries only if this opportunity is well harnessed. However, despite the increasing opportunities offered by growth of the livestock sector, smallholder livestock producers are often characterized by low levels of market participation coupled with a very low livestock market off-take rates. For example, Nkhori (2004) estimates 5% off-take rates for cattle among smallholder farmers compared to 25% in the commercial sector.

While there is a considerable amount of literature on livestock market participation, there is a dearth of information regarding to why famers move into and out of the livestock market. Several studies have attempted to understand the factors affecting smallholder decisions to participate in livestock markets for example see the work of (Costales et al 2006; Key et al 2000; Ehui et al 2003; and Lapar et al 2003). These studies highlight several recurring bottlenecks in smallholder livestock participation in markets. Poor infrastructure development such as roads and limited access to market information (prices, grades and standards) have been cited as the major limiting factors in participation in livestock markets. In many least developed countries, the road networks are poorly developed and the market information systems for most of the agricultural products tend to perform poorly. In part, this is attributed to inadequate financing and limited capacity of government agencies to collect and disseminate reliable and timely market information (Gabre-Madhin 2009). This results in information asymmetric problems which puts smallholder in a weak negotiating position especially when dealing with larger buyers or formalized channels.

The studies highlighted above provide valuable information, however, none of these studies attempted to understand the dynamics involved in participating in livestock marketing. Therefore, the current study addresses this knowledge gap about what factors influence household to move into and out of the markets or decide to participate consistently or even not to participate at all. Understanding the reasons behind these decisions may help policy makers and other development agencies working in developing countries to design interventions that deal with the problem and help reduce poverty among the rural farm households.

The cattle sub-sector in Zambia provides a useful case study for exploring the factors influencing the moving into and out of livestock markets. This paper uses a sub sample of national panel data of 524 households collected in 2001, 2004 and 2008. The 524 households are grouped into 4 different participation categories: consistent non-sellers (those who did not sell in any of the survey waves); one-time sellers (those who sold in any one of the survey waves only); two-time sellers (those who sold in any two of the three survey waves only); and consistent sellers (those who sold in all three survey waves). Because of the nature of dependent variable, we use a multinomial logit model to identify and understand the important dynamics associated with Zambian smallholder cattle markets.

The next section describes the data set used in this study. This is followed by a discussion on the underlying random utility model on which we base our empirical estimation approach used in the analysis. Results and discussions are presented in section 4, and conclusions and policy implications in the last section.

Data sources

This study draws a sub sample from nationally representative panel survey data collected from small and medium scale rural farmers in Zambia in 2001, 2004, and 2008. Though the data is old, panel data has higher leverage than cross section in tracing and explaining the dynamics of markets participation. The three waves of the survey were implemented by the Central Statistical Office (CSO) in collaboration with Michigan State University’s Food Security Research Project (FSRP) now Indaba Agricultural Policy Research Institute (IAPRI). The surveys followed the same households that were interviewed during the 1999/2000 Post-Harvest Survey (PHS). Each wave collected data on the households' cropping patterns, crop and livestock production and marketing, asset ownership, income sources, and various retrospective/current socio-demographic information on the household members. In order to understand the dynamics of market participation, we limit our sample to households that owned cattle in all three panel survey years. Of the 6,922 households interviewed in 2001, 1,217 owned cattle. Out of 1,217 households that owned cattle in 2001, 750 (62%) reported to have raised cattle in 2004 and only 524 (43%) had cattle in all the survey years. The 524 households are grouped into four different participation categories. For econometric analysis, the different participation categories were treated as alternatives without implicit order. The four participation categories in this study are

The one and two-time sellers captures those households who moved into and out of cattle markets.

Analytical framework

In this study, we use a multinomial logit model to identify and understand the important dynamics associated with Zambian smallholder cattle markets. Multinomial logit is used to model processes that involve a single outcome among several alternatives. The categorical property of the outcome (dependent) variable distinguishes the use of the multinomial logit from regression (appropriate for a continuous dependent variable), and from logit, which is appropriate for two outcomes (StataCorp. 2009).

Discrete choice decisions are frequently modeled using a random utility framework. According to the choice theory of economics, an individual i will choose an alternative j that leads to the highest utility index which is decomposed into a deterministic component and an error term (equation 1)

Uij = Xij+βj + ε ij                       (1)

where,i = 1,…, N indexes the individual decision maker, j = 0, …, J is the alternative being considered, X is a vector of observed exogenous explanatory variables, β is a coefficient vector of parameters to be estimated and  ε is an error term. An individual will choose the alternative that provides the greatest level of utility. The probability that alternative j will be chosen is given by,

Where i th be the observed outcome for the observation

The multinomial logit model predicts the probabilities of different possible outcomes of a categorically distributed dependent variable given a set of independent variables. If the error term is assumed to be independently and identically distributed (iid), the probability that the response for the i th observation is equal to the Kth outcome given J categorical outcomes is

 

 

 

However, equation 3 is unidentified as more than one solution to β that leads to the same probabilities for yi = k exits. In order to identify the model, one of the β ’s is set to zero. Setting one of the β ’s to zero ensures the adding up constraint. That is the sum of the probabilities being in the various categories must be equal to one.

This gives

 

 

 

and

 

 

 

Equation 4 estimates the probability of being in the reference group (in this study non-sellers) while equation 5 estimates the probability of being in other groups. The parameters ( β ) are estimated with the maximum log likelihood function. Equation 6 presents a reduced form of a multinomial logit.

Yij = Xijβj + εij                       (6)

Where  Yij  is the probability that household  i  chooses alternative j, β  is the vector of parameters to be estimated (of which each is different across the alternatives even though the explanatory variables are the same), X is a vector of observed exogenous explanatory variables, and  ε is an error term.

However, the coefficients (odds ratios) are not easily interpreted directly. To facilitate interpretation of results, we report the exponentiated coefficient, i.e., the relative risk ratio (RRR). Since the multinomial logit model estimates J-1 models where the jth equation is the reference group, the RRR of a coefficient measures the risk of the outcome relative to the base outcome. Variable x increases (decreases) the probability that alternative j is chosen instead of the base aalternative if RRR >1 (<1).

Factors hypothesized to influence market participation

Various factors have been identified to influence participation in previous studies. Ehui et al (2003) and Costales et al (2007) suggest that participation in livestock markets is influenced by transaction costs, human capital, physical capital, institutional factors, and financial capital. It is hypothesized that higher transaction costs discourage smallholder livestock farmers to participate in markets. Therefore, households with lower transaction costs are more likely to participate in markets since they are more likely to recover their production and marketing costs. This study uses two proxies to represent transaction costs – distance to the nearest town and a binary variable equal to one if the household is in a district along the line of rail, and zero otherwise. It is expected that households close to towns and in districts that are along the line of rail are more likely to participate in livestock markets. However, the effects may also be negative. For instance, accessibility to towns may provide better alternatives with higher pay offs than livestock marketing.

Human capital variables included in this study are age, sex, and education level of the household head and household size in adult equivalents. Education level of household head is hypothesized to increase the household’s ability to utilize market information, thereby utilizing market opportunities.

Exposure to extension services on animal husbandry was used as a proxy for institutional support. Access to extension services may translate into adoption of improved livestock production practices which could in turn increase livestock productivity. Increased livestock productivity may entail more marketable surplus. We also included a household crop commercialization index (HCI) as a proxy for marketable surplus and the extent to which the household is market oriented (Chapoto et al 2011). Households with higher HCI are less likely to participate in livestock markets as they are likely to meet the household needs through crop proceeds. In addition, to capture the effects of livestock composition on livestock market participation, we included the number of livestock owned by type. It is hypothesized that households owning small livestock are less likely to sale cattle.

To capture the influence of mortality shocks in the household, we included lagged dummy variables for death of household members in the family. Studies have shown that households experiencing mortality shocks tend to deplete their asset base which includes livestock (see Chapoto et al 2011; Muyanga et al 2013). Mortality shocks are expected to increase the likelihood for households to dispose of some of their livestock to take care of the short term funeral expenses.

We also included a binary variable for households’ social capital in the village proxy by whether the household is considered local. A household is considered local if it belongs to a clan which originally occupied the village. This may have on one hand a positive effect on the likelihood of participation due to the social network that is built within the communities. It may be assumed that social network may have greater impact in facilitating transmission of market information. On the other hand, it may reduce the likelihood as locals may be complacent about taking up entrepreneurial ventures such as selling livestock.

To account for geographical distribution of livestock and location of households and time, a set of regional and year dummies are included.


Results and discussion

Before presenting the econometric results, we present the descriptive results to provide the basic features of the data. We first present a summary of participation rates and how they evolve over time. This is followed by a discussion of the bivariate comparison of mean differences for selected attributes. Lastly, we present the econometric multinomial results.

Cattle marketing dynamics

Figure 1 presents the overall cattle market participation rates and transition into and out of cattle markets during the survey years 2001, 2004, and 2008. The results in Figure 1 show that the number of households participating in cattle markets increased over the study period. Of the 524 households that owned cattle in all the survey years, 20.2% of the households sold their cattle in 2001 and 34.8% in 2008. However, despite the increase in the number of households participating in cattle market, a greater percentage of households moved into and out of the markets during the period of study. For instance, of the 20.2% of the households who sold cattle in 2001, only 8.4% sold cattle in 2004 and 20.8% of cattle market participants in 2004 did not participate in 2001. Between 2001 and 2008, less than 5% participated consistently in cattle markets with over 40% not participating at all. Close to 55% of the households moved into and out of markets over time. The results point to the need for further analysis in order to understand the reasons behind such low levels of participation in cattle markets and the factors pushing households to sell cattle in one year but not the other. We explore these factors in the subsequent sections. We first examine the commercialization index for crop production and the various sources of income and household asset base. The assets include livestock and non-livestock.

Figure 1. Rate of participation in cattle markets
Market dynamics by income sources

The changes in crop household commercialization index (HCI) by market dynamics and year are presented in Figure 2. The results show that the HCI increased between 2001 and 2008 among the non-sellers, one-time seller and two-time seller households. In general, the results show that households that sold cattle in all the three years (consistent sellers) were less crop market oriented compared to other groups. Therefore, as the smallholder becomes more crop production oriented, they tend to move out of the cattle markets

Figure 2. Market dynamics by crop household commercialization index (HCI)

To understand these results further, we examine the sources of income, land holding size and ownership of cattle for the four groups. Results in Table 1 show that cattle non-sellers derive much of their income from crop sales. However, dependence on crop income does not necessarily mean more income as consistent sellers had about four times more household income compared to the non-sellers in all the three years. The observed increase in income in 2001, 2004 and 2008 is largely due to the sale of livestock which accounted for about 37.5%, 53.3% and 48.3% respectively. Therefore, farmers may move into or out of market depending on the level of diversification of income sources.

Table 1. Income shares by market dynamics groups

Year of
survey

Consistently
sellers

Two time
seller

One time
seller

Non-
sellers

All
households

Sample size

Weighted

2,374

9,571

22,287

24,093

58,324

Unweighted

25

96

203

200

524

Total household income per
adult equivalent in Zambian
Kwacha-2008 values

2001

1,973

1,895

1,255

1,036

1,299

2004

4,301

4,114

1,336

1,589

2,042

2008

5,087

2,634

1,448

1,088

1,640

Income share (%)

Crop

2001

36.7

54.9

67.3

76.1

67.7

2004

35.1

52.2

68.1

80.4

69.0

2008

26.7

43.2

53.7

66.4

56.1

Livestock

2001

37.5

22.8

14.8

5.18

13.1

2004

53.3

35.2

18.8

4.71

17.3

2008

48.3

32.9

26.0

10.5

21.7

Off farm

2001

25.7

22.3

17.9

18.7

19.3

2004

11.6

12.6

13.1

14.9

13.7

2008

25.0

23.9

20.3

23.1

22.3

Market dynamics by assets

A further analysis of market dynamics shows a positive association between landholding size, cattle ownership and the marketing dynamic groups (Table 2). As land holding size increases, the households move from being non-sellers to one time, two time and consistent sellers. Similarly, as the number of cattle owned increases, the households tend to move up the ladder from being non-seller to consistent sellers. Therefore, interventions that affect the herd size will have an effect on household’s movement into and out of the market. With regard to ownership of small livestock (pigs and goats), results in Table 2 show similar trend for goat ownership as that of ownership of cattle. Farmers participating in cattle markets have more goats and as the number increases, households tend to move up the ladder of market participation. However, there is no observed significant trend for pig ownership.

Table 2. Assets by market dynamics groups

Year of
survey

Consistently
sellers

Two time
seller

One time
seller

Non-
sellers

All
households

Sample size

Weighted

2,374

9,571

22,287

24,093

58,324

Unweighted

25.0

96.0

203

200

524

Land holding
size (ha)

2001

4.46

4.21

3.79

3.73

3.86

2004

3.75

4.38

3.53

3.31

3.59

2008

6.52

5.41

4.03

3.46

4.13

Number of cattle
owned (mean)

2001

29.8

14.8

11.4

6.64

10.7

2004

89.4

18.1

12.3

5.98

14.2

2008

60.2

21.5

13.7

7.50

14.2

Number of pigs
owned (mean)

2001

0.700

0.802

1.42

1.30

1.24

2004

0.316

1.38

1.55

0.948

1.22

2008

0.773

2.06

2.00

1.97

1.95

Number of goats
owned (mean)

2001

10.3

3.99

3.63

3.18

3.77

2004

14.6

5.02

4.56

4.29

4.97

2008

12.1

5.76

4.93

5.72

5.66

Basic characteristics of the four cattle market participation groups

This section summarizes the household socioeconomic characteristics of household with respect to the four cattle market dynamic group. The descriptive statistical tests are based on the initial period, 2001. The initial characteristics of individual households do influence how the household evolve over time. As shown in Table 3, cattle non-sellers had small family sizes in 2001 compared to consistently sellers (about seven compared to nine respectively). The results also show that, on average, household heads in the consistent sellers group had higher formal education than household heads in the other cattle market participation groups.

Furthermore, the proportion of households owning and selling goats is higher among the consistent sellers than among non-sellers as well as those who move in and out of the market. Pig ownership and selling were, however, more common among one and two time sellers. Analysis of variance (ANOVA) results show that differences in means are statistically significant between the groups as whole. The significant ANOVA results suggest rejecting the global null hypothesis that the means are the same across the groups being compared.

Table 3. Household initial socioeconomic characteristics by participation groups, 2001

All
households

Cattle market participation group

Consistent
seller

Two time
seller

One time
seller

Consistent
non-seller

Sig
level

Sample size

Weighted

58,324

2,374

9,571

22,287

24,093

Unweighted

524

25

96

203

200

Demographics

Female headed HH (%)

7.99

8.50

6.42

9.94

6.76

***

Mean number of HH members

7.91

9.29

8.72

7.97

7.40

***

Mean age of HH head (years)

49.5

48.4

49.2

49.3

49.94

***

Mean years of schooling of HH head

6.18

8.24

6.88

6.20

5.67

***

Assets

HH owns goat(s) (%)

40.3

68.2

32.2

40.4

40.8

***

HH sold goat(s) (%)

14.9

41.2

17.6

16.1

10.1

***

HH owns pig(s) (%)

21.8

18.7

22.7

22.6

21.1

***

HH sold pig(s) (%)

5.06

0.88

5.74

6.29

4.06

***

Mean landholding size (ha)

3.86

4.46

4.21

3.79

3.73

***

Note: HH = household. significance level *** p<0.01

However, ANOVA is limited because it does not indicate which of the means of the four groups is different from others. Thus, a post hoc test was performed to indicate whether the observed differences were statistically significant. An example of the post hoc test results is shown in table 4. While table 3 show significant differences across all the groups for female headed household, for example, table 4 show insignificant differences between non-sellers and two time sellers as well as between one time sellers and consistently sellers.

Table 4. Mean differences Post hoc test (Tukey HSD)

Dependent
Variable

(I) `four groups of
households selling cattle'

(J) `four groups of
households selling cattle'

Mean Difference
(I-J)

Std.
Error

Sig.

Female headed
household

consistently non sellers

one time seller

-0.032***

0.003

0.000

two time seller

0.003

0.003

0.730

consistently sellers

-0.017**

0.006

0.015

one time seller

consistently non sellers

0.032***

0.003

0.000

two time seller

0.035***

0.003

0.000

consistently sellers

0.014

0.006

0.066

two time seller

consistently non sellers

-0.003

0.003

0.730

one time seller

-0.035***

0.003

0.000

consistently sellers

-0.021***

0.006

0.004

consistently sellers

consistently non sellers

0.017**

0.006

0.015

one time seller

-0.014

0.006

0.066

two time seller

0.021***

0.006

0.004

significance level *** p<0.01, ** p<0.05, * p<0.10

Econometric results

Table 5 presents the econometrics results. As discussed under methods, the RRR of a coefficient measures the risk of the outcome relative to the base outcome. Consistent non seller is the base group in this study. A unit increase in variable x increases (decreases) the probability that alternative j is chosen instead of the base alternative if RRR >1 (<1). We begin our discussion with the human capital factors. This is then followed by income sources and assets, market access, social capital and institutional factors and lastly the effects of household shocks.

Human capital

The multinomial logit results in Table 4 confirm many of the hypotheses emerging from the descriptive analysis. The results show positive association of human capital variables with the likelihood of households moving into the market. Important is the education variable which is highly significant. In general, one more year of schooling increases the probability of being in one of the groups relative to non-participation. This finding highlights the importance of education in increasing the ability of households to utilize, analyze and comprehend market information and thereby utilizing market opportunities Therefore, investing in education can positively affect the livestock sector.

Table 5. Multinomial regression results

Explanatory variables

Relative Risk Ratio (RRR)

Consistently
sellers

Two-time
seller

One-time
seller

Household human capital

Female headed HH (=1, 0 otherwise)

2.37*

1.91**

1.33

(1.12)

(0.556)

(0.321)

HH members (adult equivalent)

1.05

1.08***

1.06***

(0.0374)

(0.0259)

(0.0222)

Age of HH head (years)

1.02

1.016**

1.00

(0.0109)

(0.0064)

(0.00500)

Years of schooling of HH head

1.130***

1.10***

1.054***

(0.0441)

(0.0265)

(0.0210)

Income sources and assets

Crop household commercialization Index

0.468

0.534**

0.995

(0.253)

(0.167)

(0.247)

One or more HH members engaged in off
farm employment(=1, 0 otherwise0

1.91**

1.19

1.19

(0.609)

(0.210)

(0.166)

landholding size (ha)

1.04*

1.031**

0.999

(0.0197)

(0.0153)

(0.0146)

Number of goats owned

0.935

0.997

1.009

(0.0426)

(0.0222)

(0.0165)

Number of pigs owned

1.05***

1.02**

1.02*

(0.0129)

(0.0100)

(0.00890)

Shocks

Recent death in HH (=1, 0 otherwise)

0.803

1.18

1.15

(0.324)

(0.258)

(0.203)

Number of cattle that died due to illness

1.14***

1.12***

1.13***

(0.0340)

(0.0328)

(0.0319)

Institution factors

HH received market information (=1, 0 otherwise)

1.04

0.866

0.774*

(0.334)

(0.163)

(0.116)

HH received extension service(s) (=1, 0 otherwise)

1.22

1.56**

1.52**

(0.421)

(0.312)

(0.248)

Market access

Kilometers to nearest main road

0.953

0.976

1.04***

(0.0420)

(0.0196)

(0.0106)

Household social capital

HH considered local (=1, 0 otherwise)

1.47

0.681*

0.671**

(0.611)

(0.152)

(0.126)

Agro ecological zone (Region II is the base)

Region I: low rainfall (less than 800 mm),

2.81**

3.14***

1.99**

(1.26)

(0.939)

(0.535)

Region III: high rainfall (over 1000 mm)

0.0000

4.47***

3.15***

(0.00120)

(1.53)

(0.978)

Year 2004

1.07

1.15

1.07

(0.439)

(0.275)

(0.207)

Year 2008

0.756

0.892

0.945

(0.252)

(0.174)

(0.146)

Constant

0.00630***

0.0526***

0.300***

(0.00590)

(0.0279)

(0.122)

Notes: Standard errors in parentheses; significance level *** p<0.01, ** p<0.05, * p<0.1.
HH = household. N=1286 panel households for all models

Income sources and assets

The household crop commercialization index is a proxy for the household’s degree of crop commercialization. The results in figure 2 are supported by the econometric results in Table 4, which show that crop commercialization dampens the likelihood of household moving into cattle markets. Thus, a one percent increase in HCI results in households moving out of the market and being in a non-seller group. Thus livestock only cushions the financial stress in periods of less crop production. Results also show that households selling cattle do also engage in off-farm activities. More specifically, we find that a positive relationship between participation in off-farm activities and households’ likelihood to move into cattle markets. Generally, the positive relationship between off-farm activities and livestock market participation seem to suggest that engaging in off-farm activities provide greater opportunities to interact and build market relationships with would-be buyers of cattle on one hand. On the other hand, literature on the linkages between nonfarm sector and farm sector highlights the importance of off-farm income in the development of farm enterprises and vice versa. Specifically, income earned from off-farm activities can benefit farm activities through financing farm activities and investment in increasing farm productivity (Dorward et al 2004). The descriptive results in table 2 are supported by the econometric results in Table 4, which show significant effects for two time and consistent sellers. The coefficient suggests an enhancing effect. These results support Turner (2004) who found land to be an important asset that supports the production of livestock.

Market access, social capital and institutional factors

Using the distance to the nearest main road as a proxy for accessibility to markets and household considered local as a proxy for access to local institutions, we examine the relationship between rural households’ participation in cattle markets and market access. The descriptive results in table 2 are supported by the econometric results in Table 4, which show significant effects of land holding size for two time and consistent sellers. This suggests the importance of the transport sector in cattle marketing. On the other hand, the results show that household considered local are likely to move out of the market. This may indicate complacent among local individual who would not take up entrepreneurial activities.

We further examine the institutional factors using access to market information and extension services. Access to market information, though not significant, is likely to increase the likelihood of households being a consistent seller. However, this depends on the market information received. In most cases farmers have access to maize marketing prices and this explains the significant coefficient of market information on the one time sellers. Likewise, access to extension service has positive effect on the likelihood of household to move into the markets as indicated by a greater than one significant coefficient.

Household shocks

If a household experiences the death of any member, they are more likely to move into the market though the coefficient is not statistically significant. Studies on poverty dynamics have shown that chronic illness and death tends to drain households assets such as livestock to cater for medicines, medical bills, caretaking and funeral expenses (see Muyanga et al 2010; Chapoto et al 2011). Cattle deaths due to disease are also likely to increase the probability of moving into the market. A possible explanation for this effect is that during disease outbreaks of national importance government through the veterinary department restricts movement of livestock into and out of affected areas. .In response, households often slaughter and sell animals through informal channels to avoid incurring further losses. Though positive on market participation, this result suggests the need to reinforce health service delivery in order to contain disease outbreaks.


Conclusion and suggestions


Acknowledgements

The author wishes to acknowledge Indaba Agricultural Policy Research Institute for the data used in this study.


References

Chapoto A, Banda D, Haggblade S and Hamukwala P 2011 Factors affecting poverty dynamics in rural Zambia. (No. 109888). Michigan State University, Department of Agricultural, Food, and Resource Economics.

Costales A, Delgado C L, Catelo M A, Lapar M, Tiongco M, Ehui S K and Bautista A Z 2006 Scale and access issues affecting smallholder hog producers in an expanding peri-urban market: Southern Luzon, Philippines (Vol. 151). International Food Policy Research Institute.

CSO/FSRP Post Harvest Survey 1999/2000 and Supplemental Surveys 2001, 2004, and 2008. Lusaka: FSRP.

Delgado C L 2003 Rising consumption of meat and milk in developing countries has created a new food revolution. The Journal of Nutrition, 133.11:3907S-3910S. http://jn.nutrition.org/content/133/11/3907S.full.pdf+html

Delgado C L, Rosegrant M W, Steinfeld H, Ehui S K, and Courbois C 1999 Livestock to 2020: The next food revolution (Vol. 61). International Food Policy Research Institute. http://www.ifpri.org/publication/livestock-2020

Dorward A , Fan S, Kydd J, Lofgren H, Morrison J, Poulton C, Rao N, Smith L, Tchale H, Thorat S, Urey I and Wobst P 2004 Institutions and policies for pro-poor agricultural growth, Development Policy Review, 22.6: 611-622.

Ehui S, Benin S, and Paulos Z 2003 . Policy options for improving market participation and sales of smallholder livestock producers: A case study of Ethiopia. Paper Submitted At the 2nd International Conference on Contemporary Development issues Addis Ababa, July, 11-13, 2003. http://scholarworks.wmich.edu/cgi/viewcontent.cgi?article=1057&context=africancenter_icad_archive

Gabre-Madhin E 2009. A market for all farmers: market institutions and smallholder participation. CEGA working paper series No. AfD-0903. Berkeley, CA: University of California Center of Evaluation for Global Action.

Key N, Sadoulet E and de Janvry A 2000 Transactions costs and agricultural household supply response. American Journal of Agricultural Economics 82.2: 245–259.

Lapar M L, Holloway G, and Ehui S 2003 Policy options promoting market participation among smallholder livestock producers: a case study from the Philippines. Food Policy 28:187–211.

Livestock data innovation in Africa 2014 Business and livelihoods in African livestock investments to overcome information gaps. Washington, DC and Rome: The World Bank and the Food and Agriculture Organization of the United Nations. http://www.fao.org/3/a-i3724e.pdf.

Muyanga M, Jayne T S and Burke W J 2013 Pathways into and out of poverty: A study of rural household wealth dynamics in Kenya. The Journal of Development Studies, 49.10:1358-1374. http://www.tandfonline.com/doi/pdf/10.1080/00220388.2013.812197

Nkhori P A 2004 The impact of transaction costs on the choice of cattle markets in Mahalapye district, Botswana (Doctoral dissertation, University of Pretoria). http://repository.up.ac.za/handle/2263/26363

Reardon T, Tschirley D, Dolislager M, Snyder J, Hu C, White S 2014 Urbanization, diet change, and transformation of food supply chains in Asia. East Lansing, MI: Global Center for Food Systems Innovation.

Stata Corp LP 2009 Stata statistical software release 11. Stata Press Publication.

Turner, R L 2004 Livestock production and the rural poor in Andhra Pradesh and Orissa states, India. Pro-poor Livestock Policy Initiative Working Paper No. 9. Rome: FAO, PPLPI.


Received 8 December 2015; Accepted 28 February 2016; Published 1 April 2016

Go to top