Livestock Research for Rural Development 24 (10) 2012 Guide for preparation of papers LRRD Newsletter

Citation of this paper

Poultry production and rural poverty among small-scale farmers in Mzimba District of Malawi

Maganga M Assa

Department of Agricultural and Applied Economics, Bunda College of Agriculture, P.O. Box 219, Lilongwe, Malawi


This paper evaluates the role of participating in poultry production on household income and rural poverty in Mzimba district, Malawi. The study utilizes cross-sectional farm level household data collected in 2011. The paper computes income-based poverty measures and investigates their sensitivity to the use of different poverty lines. Robust poverty comparisons across the poultry and non-poultry farmers reveal that poverty is in fact higher for the non-poultry compared to the poultry farmers. Thus, participating in poultry production has a significant positive impact on household income and poverty reduction.

Keywords: FGT, Poverty, Small-scale poultry production, stochastic dominance


Poultry production is the fastest growing component of global meat production, with developing and transitional economies taking a leading role. In addition to providing opportunities to increase poultry exports, rising poultry production spurs growth in global import demand for feeds and other inputs and generates up- and downstream investment opportunities (Regmi 2001). 

World poultry meat output increased nearly eightfold between 1961 and 2001, while the output in middle-income countries even rose more than twelve fold. In 1961, middle-income countries produced 34 percent of world poultry meat, high-income countries 61 percent, and low-income countries the remaining 5 percent. By the mid-1990s, middle-income country production had reached a level of 47 percent of the output of high-income countries. By 2001, middle-income countries accounted for the major share of world poultry production (52 percent) compared with 42 percent in high-income countries and less than 6 percent in low-income countries (Regmi 2001). However in Malawi, the trends in levels and growth of livestock per capita show that poultry production has been declining. The numbers of chickens per capita have been declining, with the average in the last past five years being lower than that recorded in the early 1970s. Despite this decline, agricultural policy documents specify several interventions to boost poultry production among small scale farmers in a conduit of poverty alleviation (ASWAP 2011). 

The objective of poverty and malnutrition alleviation cannot be pinned down by a single peg. No single effort will achieve a major impact in isolation. However, poultry has shown to offer a practical and micro level step in alleviation of rural poverty. There is evidence that investments in small scale poultry farming generate handsome returns and contribute to poverty reduction and increased food security in regions where a large share of the population keeps some poultry birds (Jensen and Dolberg 2003; Mack et al 2005; Pica Ciamarra and Otte 2010).  In Malawi, about 83% of rural households are estimated to keep flocks of 1 to 20 birds (Gondwe and Wollny 2002). 

Although there is no universal definition of poverty, everyone seems to agree that it exists when one or more persons fail to attain a level of well-being deemed to constitute a reasonable minimum by the standards of that society (Ravallion 1993). This situation , which has been ascribed in some quarters to production failure owing to a suppression of markets and in some other quarters to distributional failure (Dasgupta 1998), is characterised by disease, low life expectancy, and physical and mental retardation. Therefore, for any effective plan to reduce poverty, the poverty dynamics of the population has to be understood. 

Area specific empirical evidence on poverty dynamics among small scale farmers and how poultry production affects poverty is imperative for appropriate policy choices, program or reform management towards welfare shift in rural agrarian economy like Malawi. This evidence is brought on surface by estimating the dimensions of poverty among small scale farmers and determining the impact of poultry production on poverty among small scale farmers in Mzimba district. 


Study area and data

Data regarding various components of the small holder’s poultry production were generated through a field survey study. The survey was conducted in the rural areas of Mzimba District of Northern part of Malawi during the month of May-June 2011. A semi-structure survey questionnaire was developed to collect information and a total of 89 farmers located in 8 villages were interviewed. The sample was distributed across 7 Agricultural Extension Planning Areas (EPAs). These EPAs were selected using simple random sampling. In the second stage, villages were selected using systematic sampling and finally in each village household were sampled proportionate to size sampling procedure. 67 of these households engaged in poultry farming with the remaining 22 did not engage in poultry farming. Both descriptive and quantitative analyses were carried out in the study. The descriptive analysis including means and percentage analysis were used to describe the socio-economic and demographic characteristics of farmers and their households in the study area.

Poverty analysis

To present the poverty profile of the farmers in the study area, various methods were used to estimate the extent and manifestations of poverty among the sample farmers. Specifically, the headcount, poverty incidence and poverty gap measures were employed in the analysis. Headcount measure estimates the absolute number of the poor in the sample. Poverty incidence estimates the percent of the poor in the total sample. Poverty gap measures the intensity of poverty based on the extent of income shortfalls below the poverty line by the poor in the sample (Olubanjo 1998). Specifically, the paper uses Foster-Greer-Thorbecke (FGT) which measures and subsumes the headcount index and the poverty gap, and provides the distributional sensitive measure through the choice of a poverty aversion parameter “α”; the larger the value of the “α”, the greater the weight given by the index to the severity of poverty (Anyawu 1997). The general specification of the model is given below:


Where, α is a non-negative parameter; Yi is income per person in the ith household; is the number of households below the poverty line; xis the poverty line value or threshold value of income; and N is the number of persons in the sampling population. α = 1 for poverty incidence, 2 for poverty depth and 3 for severity.  

and α is the number of persons in the sampling population. α = 1 for poverty incidence, 2 for poverty depth and 3 for severity.

To compare poverty levels for poultry farmers and non-poultry farmers, it is worth to compare their income levels and check if distribution of income in one group always dominates the other. To implement this procedure, we use stochastic dominance algorithm. Stochastic dominance tests in poverty analysis checks whether the poverty ordering remains the same over a variety of poverty lines, based on the comparisons of cumulative distribution functions (CDF). 

Consider two distributions of welfare indicators with cumulative distribution functions, FA and FB , with positive definite real numbers. Let




Econometric construct

To measure the effect of poultry production on poverty, an endogenous switching regression was employed which is able to take care of selection biases.  We specify the selection equation for poultry production as

Where, Gi * is the unobservable or latent variable for technology adoption, Gi is its observable counterpart (the dependent variable participation in poultry production equals one, if the farmer adopts and zero otherwise), Xi  are non-stochastic vectors of observed farm and non-farm characteristics determining participation and ui is random disturbance associated with the participation in poultry production and and β is a vector of parameters to be estimated.

Participation in poultry production by a farmer is assumed to be derived from the maximization of a discounted expected utility of benefits subject to farmer specific characteristics. We hypothesise that a vector of household specific variables influence the choice of participation. 

To account for selection biases we adopt an endogenous switching regression model of welfare outcomes, (i.e. household income) where farmers face two regimes (1) to participate, and (2) not to participate defined as follows:


Where Yi is household income in regimes 1 and 2, Ji   represents a vector of exogenous variables thought to influence household income. The error terms in Eq. (6), conditional on the sample selection criterion, have non-zero expected values (Lee 1978; Maddala 1983). Lee (1978) treats sample selection as a missing-variable problem.  

The error terms are assumed to have a tri-variate normal distribution with zero mean and non-singular covariance matrix specified as;


Full information maximum likelihood (FIML) estimation provides an efficient method to estimate endogenous switching regression models (Lee and Trost, 1978; Lokshin and Sajaia 2004). Given the assumption of trivariate normal distribution for the error terms, the logarithmic likelihood function for the system of equations (6) and (7 and 8) can be given as

Empirical findings and discussion

Socio-demographic characteristics of sample farmers in Mzimba district are presented in Table 1.  Analysis has revealed that majority of households (64%) are headed by males while 36% are female headed. Average age (49years) of farmers did not differ between poultry and non poultry farmers. Distribution of respondents with respect to educational status reveals that 97% of them attained different levels of formal education. An average household earn MK 126,516 for poultry farmers and 12,978 for non poultry farmers. . Frequency of extension visits, membership to farmer organization, livestock training and access to credit were most common among poultry farmers than non poultry farmers (Table 1). There are more farmers who are aware of climate change among poultry farmers. Probably this could be one of the reasons they engage in poultry production as they know that poultry may provide an avenue for escaping climate change impacts. The results reveal that family sizes are larger among adopter than non adopters.  

The housing conditions of a household provide good indicator of welfare measurement. Among poultry farmers, about 69% of farmers live in grass thatched houses and only 31% live in houses with iron sheets. About 84% lived in housed constructed from burnt bricks with 16% in mud houses. Among non poultry farmers, about 57% of farmers live in grass thatched houses and only 43% live in houses with iron sheets. About 78% lived in housed constructed from burnt bricks with 22% in mud houses.

Table 1: Definition of variables and descriptive statistics         



Poultry farmers

Non poultry farmers





Land size




Household income




Extension visit

1 =yes



Membership to organization




Access to livestock information




Livestock training




Household size

No of persons



Credit status

1=access to credit



Distance to main market




Gender of household head

1= Male



Average price of chicken (buy/sell)




Household size

No of persons



Climate change awareness




Knowledge of poultry drug









lower primary




upper primary




junior secondary




senior secondary











Housing condition

Grass thatched




Iron sheet








Burnt bricks



As shown on Table 2, poverty incidence among small scale farming families was found to be 64% resulting from 60% of household involved in poultry production and 85% of household without poultry. This implies that, overly, 60% of the respondent farmers and 85% of non poultry farmers were actually poor. This proportion invariably represents the poverty incidence among the sample. Those who are involved in poultry production reported lower than 77% as at 2002 reported by Chirwa (2005). 

The poverty-gap index (PG), defined by the mean distance below the poverty line as a proportion of that line, is usually interpreted as a measure of poverty depth. The poverty gap of the sample was 42%. The figure was 38% for poultry farmers and 56% for non-poultry farmers. This implies that poor poultry farmers required 38% and non-poultry farmers required 56% of the poverty line to get out of poverty.

Table 2: Foster-Greer-Thorbecke (FGT) class of poverty measures for the study sample

FGT Measures

Poultry farmers

Non-poultry farmers


Incidence of poverty




Depth of poverty




Severity of poverty




Computed from 2011 study data

Finally, overall severity of poverty computed by FGT was 33%. This measure implies that a distinction is made between the poor and the poorest. This follows since the poverty depth is not sensitive to re-distribution among the poor. The assumption with the poverty gap is that a Kwacha gained by the poor would have the same effect on poverty as that gained by the moderately poor farmers. As such, to capture the sensitivity to income re-distribution among the poor and non-poor, there exists the need to estimate the severity of poverty among the study sample. There was a difference between poverty severity of poultry farmers (30%) and non-poultry farmers (42%).  

The stochastic dominance tests show a similar result. The test statistics are calculated at each value of the poverty line, where we only considered 20 poverty lines between MK676 and MK3800 per capita per year. The estimated headcount ratios along with the t-statistics of the difference for each of 20 points were obtained. Figures 1 shows the poverty incidence curve. A close inspection of the poverty incidence curve (Figure 1) reveals that there is first-order dominance because the CDF of adopters is always to the right of non-adopters but results of the test statistic for each value is insignificant indicating there is no first-order dominance. Given that first order dominance is observed, it left no desire to test for higher order dominances. 

That is, poverty as measured by head count index is unambiguously lower for poultry farmers than for those who do not engage in poultry farming, regardless of the poverty line chosen. This result underscores the role of livestock in contributing to poverty reduction through increasing per capita household income. This suggests that the poultry production had a measurable impact on reducing the incidence of poverty.

Figure 1. Poverty incidence curves

The full information maximum likelihood estimates of the endogenous switching regression model that can control for unobservable selection bias are reported in Table 2 and adoption of poultry production in Mzimba. The first column presents the estimated coefficients of selection equation (2) on adopting poultry production or not whereas the second and third column presents the household incomes functions for adopters and non adopters. 

The Full Information Maximum Likelihood estimates of the endogenous switching regression model of household welfare are presented in Table 2. The last but one row gives the estimates of the coefficients of correlation between the random errors in the system of equations. The estimated coefficient of correlation between the adoption equation and the adopters’ welfare function   is positive and significant. The adoption model results and the switching regression model results together suggest that both observed and unobserved factors influence the decision to adopt poultry production and dynamics of their welfare given the adoption decision. The significance of the coefficient of correlation between the adoption equation and the welfare function for adopters indicates that self-selection occurred in the adoption of poultry production. That is, (1) poultry production had a significant impact on household welfare among adopters; and (2) adopters would have got greater benefits from improved cowpea varieties than non-adopters, had non-adopters chosen to adopt. However, the estimated coefficient of correlation between the adoption equation and the non-adopters’ welfare function, , is not significantly different from zero, implying that adopters and non-adopters operate on same indifference curve in absence of poultry production, given their observed characteristics. The initial differences between adopters and non-adopters, though insignificant, brought about differential effects of poultry production on the two groups, confirming the sensitivity of poultry impacts to initial differences due to unobserved factors.


Table 3: Full information maximum likelihood estimates of the switching regression model Dependent variable: Poultry production and household annual income for Mzimba, Malawi


FIML Endogenous Switching Regression

Poultry participation (1/0)

Participation =1

poultry farmers

Participation =0

Non poultry farmers

Age of household head

0.01 (0.002)***

0.54 (0.33)

0.58 (0.35)*

Gender of household head

0.62 (0.52)

1.73 (0.70)**

0.17 (0.35)

Lower primary (std 1 - 4)



0.16 (0.07)**

Upper primary (std 5 - 8)



0.13 (0.05)**

Junior secondary (form 1 - 2)

-0.42 (0.18)**


1.32 (1.13)

Senior secondary (form 3 - 4)

-0.5 (0.21)**

-0.01 (0.04)

0.61 (5.11)

Tertiary, diploma

0.67 (0.52)

-0.001 (0.003)

0.05 (0.14)

Had livestock extension services

0.25 (0.17)


2.33 (4.65)

Total Land size

0.42 (0.18)**

1.48 (0.41)***

0.88 (0.05)***

Average price of chicken (buy/sell)

-0.002 (0.002)

-1.89 (0.54)***

0.66 (0.26)**

Household size

0.26 (0.43)

0.02 (0.05)

0.43 (0.17)**

Access to market information

-0.67 (0.27)**


-1.99 (2.13)

Climate change awareness

0.02 (0.01)***



Access to credit

64.8 (92.3)



Log of distance to the market

1.12 (0.27)***



Member to farmer organization

0.18 (0.06)***



Knowledge of poultry drug

0.39 (0.14)***




2.02 (0.65)***


6.53 (11.6)

LR test of independence of equations (χ2)





Wald χ2 (12)   





0.89 (0.06)***

0.892 (0.06)***


0.63 (0.14)***

-0.21 (0.44)

*,** and *** mean significant at 10%, 5% and 1%, respectively

Figures in parenthesis are std. errors.

Conclusion and policy implication


Anyawu JD 1997 Poverty in Nigeria: Concepts, Measurements and Determinants. In: Obaseki, PJ (Eds.), Poverty Alleviation in Nigeria. Proceedings of Nigeria Economic Society Annual Conference. 

Chirwa EW 2005 Macroeconomic Policies and Poverty Reduction in Malawi: Can we Infer from Panel Data. GDN. 

Dasgupta P 1998 The economics of poverty in poor countries. Scandinavian Journal of Economics 100(1): 41-68. 

Davidson R Duclos JY 2000 Statistical inference for stochastic dominance and for the measurement of poverty and inequality. Econometrica 68(6), 1435–1464. 

Foster  JE and Shorrocks AF 1988 Inequality and poverty orderings. European Economic Review. 32(2-3), 654–661 

Gondwe TNP and Wollny C 2002 Challenges to Organic Farming and Sustainable Land Use in the Tropics and Subtropics. State of Poultry Production and Breeding in Malawi. Deutscher Tropentag Witzenhausen, October 9-11, 

Jensen HA and Dolberg F 2003 A Conceptual Tool for Using Poultry as a Tool in Poverty Alleviation. Revised version. Paper delivered at the International Conference on Staying Poor: Chronic Poverty and Development Policy, IDPM, University of Manchester. 

Lee  LF and Trost RP 1978 Estimation of some limited dependent variable models with application to housing demand. Journal of Econometrics 8: 357-382. 

Lokshin M and Sajaia Z 2004  Maximum Likelihood Estimation of Endogenous Switching Regression Models. Stata Journal 4(3): 282-289. 

Mack SD Hoffmann and J Otte 2005 The Contribution of Poultry to Rural Development. World’s Poultry Science Journal, 61(1): 7–14. 

Maddala  GS 1983 Limited dependent and qualitative variables in econometrics. Cambridge, U.K.: Cambridge University Press. 

Olubanjo OO 1998 Profile of the poor and poverty alleviation in the countryside: The case of four spatial settlements in Ogun State, Nigeria, The Nigerian Rural Sociologist 2(1): 31-40.

Pica Ciamarra U and Otte J 2010 Poultry, Food Security and Poverty in India: Looking beyond the Farm‐gate. World’s Poultry Science Journal, 66(2): 309–320. 

Ravallion M 1993 Poverty comparisons. Fundamentals in pure and applied economics, Vol. 56.  Chur, Switzerland: Harwood Academic Press. 

Ravallion M 1994 Poverty Comparisons. Harwood Academic Press, Chur, Switzerland 

Regmi A (Ed) 2001 Changing Structure of Global Food Consumption and Trade [Online]. Washington DC [USA]. Economic Research Service/USDA. Pp 103-107. Accessed 2012 July 24.

Received 8 August 2012; Accepted 3 September 2012; Published 1 October 2012

Go to top