Species diversification, livestock production and income of the poor in the Indian State of Andhra Pradesh

S Akter, J Farrington, P Deshingkar, L Rao* and A Freeman**

Overseas Development Institute, London
aktshahe@aol.com
^*Overseas Development Institute, India
^**International Livestock Research Institute, Nairobi, Kenya

Abstract

This paper investigates the factors associated with private sector smallholder livestock and the associated income of the poor in the Indian States of Andhra Pradesh. The data used are drawn from the Livelihood Options Study led by the Overseas Development Institute, London, specifically the Census Survey 2001/2 and Panel Survey of early 2005. Poorer households depend disproportionally on livestock.

Important changes were noted; the average size of any single species declined considerably in the five year period beginning from 1996/7, the number of farms keeping the species rose in the 1996/7 to 2001/2 period but dropped in the recent 2001/2 to 2003/4 period. The poorer households kept fewer small ruminants and poultry in the latter period but richer households kept more. This might suggest that the technology is shifting towards what the better off can afford like the intensive poultry keeping, and that they have better access to certain resources such as crop residues. On the other hand, policies such as the ban on grazing by goats will also have affected the poor more since they rely on open-grazing methods.  Farmers identified shock/stress variables and natural environment as important for the decrease in livestock population. Livestock act as a very real means of smoothing income by allowing debts to be repaid, farm inputs and medical treatment to be purchased, and dowry to be paid. More than 22% of the farmers mentioned disease problems as a cause of the decline in livestock population. This problem needs to be addressed, possibly through public-private partnership arrangements as are being tried in other countries.

Key words: India, poverty,, smallholder livestock, species diversification

Introduction

The livestock sector is valued as one of the main global drivers of agriculture as well as one of the sectors having enormous potential to poverty reduction (FAO 2005, Holmann et al 2005). Its growth in recent years has been high especially in developing countries, where annual growth rates in the last ten years in livestock have been 3.77 percent compared to 2.71 percent in crops and 1.18 percent in non-food commodities. Small scale production, especially by the poor, is heavily dependent on common pool resources (CPRs), the protection of which is very important from environmental perspectives. 

Public policies in many countries are directed towards protection resulting in changes in the livestock composition in terms of diversification of species for both large and small ruminants. In the shorter term – when restrictions are most strict - households may become more dependent on crop residues in the face of diminishing accessibility of CPRs in grazing systems (Turton 2000). For this reason, the income of the poor who have less access to crop residues are expected to reduce and the poor are expected to end up with decreased livestock production. However in the longer term, better utilisation of CPRs via e.g. watershed development can generate important fodder resources for cut and carry systems (Farrington and Lobo 1997). This may again enhance the income of the poor from livestock if they could survive until then but may change the composition of species due to the change in the feeding system.

This paper investigates the factors associated with livestock keeping, changes on the species composition and how this composition is affecting the income of the poor in the Indian State of Andhra Pradesh. It draws inter alia on arguments and evidence generated during ongoing ODI work (Deshingkar et al 2006).
 

Data and methodology

The data used for this paper were drawn from the Livelihood Options Study led by the Overseas Development Institute, London, specifically the Census Survey 2001/2 and Panel Survey of early 2005. The project carried out several rounds of survey in the State of Andhra Pradesh (AP) comprising a census of 4767 households having household characteristics and asset information and a panel survey of 353 households with more information on changes in the livestock species, reasons for change and contribution on household shock. The census survey was done when drought conditions had just begun and the 2005 round was done at the end of a drought. The survey selected three districts purposively out of twenty-three from three different regions considering the differences in terms of agro-ecological and other characteristics (Farrington et al 2006). The regions were Rayalseema, Coastal Andhra and Telangana, and the selected districts were Chittoor, Krishna and Medak from the respective region. From each district two contrasting villages were chosen, based on a number of different criteria including proximity to urban areas, roads and markets; social and economic indicators of development; absence of factionalism and extremism; coverage by pro-poor programmes etc. The villages in the Chittoor district were Oteripalli (OP) and Voolapadu (VP), in the Krishna district were Kosuru (KO) and Kamalapuram (KA), and in the Medak district were Gummadidala (GU) and Madhwar (MD). 

About 40% of the census households were keeping one or more types of livestock species in 2001/2. Of the panel households 52% were raising livestock in 2001/2 but this was 2 percentage points less by 2003/4. The nearer the village from a town in a district (KO and GU), the less intense was the farming; exceptions were the Chittoor district villages, OP (near) and VP (far) having an opposite pattern. VP is situated in the dry part of the district with specialization in sericulture and has more labour market linkages with construction sites.  Over the years, a sharp decline in livestock population was observed in this village due to drought, declining access to CPRs as a result of privatisation of commons, reduction in the mulberry cultivation, foot and mouth disease, and a higher scale of migration and tractorisation. Problems such as drought, privatisation and elite capture of commons were also prevalent in OP but a dairy technology based on stall feeding of crossbred cows was more widely adopted here and so the decline in livestock farming was not as serious as in VP.  

Average holding among livestock keeping households was very low in the survey areas; on average number around 22 with a minimum of 1 to a maximum of 10004, and skewed, with around half of them keeping 1-2. In standardised units it was 1 SLU with a range of 0.02-204 (SLU is defined in Table 1. Age/size difference was not taken into account in this standardization due to lack of data). Average size based on livestock keeping households (1845) was 2.4 SLU). Only 0.1% of the households kept more than 18 SLU (6 households out of 4647, they were commercial poultry farms with flock size ranging from 1000 to 10000). 

In this study livestock keeping decisions are postulated to depend on household level characteristics, land, income, assets and livelihood diversification and village level factors such as market linkages, access to common pool resources, infrastructure and institutions. Under these premises we used multivariate regression analysis to examine the relationship between livestock keeping and associated factors. Tabular analysis based on univariate income quintile class was applied on panel data to examine the changes in species composition, reasons for the change and contribution to income.

 

Livestock ownership and associated variables

We specified the ownership equation taking SLU per adult equivalent as the dependent variable and independent variables as defined in Table 1. Livestock ownership is a type of output variable, so that this equation should be a production function and should ideally include input variables such as feed, veterinary care, housing and extension services. These variables were omitted due to lack of data.

The model was estimated by ordinary least squares method for the full sample as well as for different quintile groups (estimated models are presented in the Appendix) (Full sample in this case comprises 4556 households, for other households income data were not collected or were missing.). F-tests suggested separate regressions and Hausman test for simultaneity between income and livestock units suggested that the problem was not serious. The independent variables gave highly significant explanations of the variation in ownership per adult equivalent. The log transformation was conducive to the resolution of the unequal variance problem, and to easy interpretation of the results. The elasticities of livestock ownership per adult equivalent for the respective variables are shown in Table 2.

In the semi-subsistence setting income is more dependent on livestock than the other way round. This is shown by the poor performance of the income variable in the estimated equations reported in Table 3.  The income variable was not significant in 3 equations out of 5 estimated. The signs were consistent.

The important variables that determine livestock ownership were land, diversification and species (Species, migration and some other variables were measured by dummy variables and so the coefficients were not included in Table 2 but in Table 3). Assets, landlessness and migration were important determinants. Village specific factors had stronger influence on livestock keeping. The status as scheduled caste strongly reduced the ownership.

The variable species was responsible for the most of explanatory power of the models. On an average, households owning a single species had lower numbers of livestock, suggesting that few had specialized, and that the majority pursued risk aversion strategies of keeping several different types of livestock. Qualitative assessment by Deshingkar et al (2006) identified a number of barriers to specialisation, including inadequate input supply. In the 2005 panel survey we noted the increase/decrease in the livestock types and the associated major reasons and will be discussed in the subsequent sections. The size of the species coefficient was higher for the poorer households than the better-off, implying that the poorer were more risk averse.

The richer households responded relatively more with respect to land. Poorer households which had migrated member had much less ownership than the richer households with migration. The impact of livelihood diversification on livestock ownership was higher for the poorer households, because poorer households relied more on livestock than the richer to increase local livelihood diversification, but for those migrating, livestock were required only in a few cases such as for hauling carts for sugar cane harvesting, and were otherwise more of a hindrance than a help in diversification through migration.
 

Changes in the species composition

Over the five year period beginning 1996/7, intensive technology appeared supportive only for chicken/ducks. The maximum number had fallen for almost all species (Table 4).

Farming density had increased for all types except for pigs. Some of the increase was due to population growth but most of the change in dairy, bovine, goat, poultry and duck might be due to some adaptive mechanism under the changing environment. Most of the farms, in particular the poorer groups diversified the holding by decreasing the size of a single species. The trade off between the ownership of a species and diversification failed to increase in the total herd size, which reduced by about 8% over this period. Thus on an average, the drop in one species at farm level more than offset the rise in other species.

Most of the large livestock farms, which increased their total herd size over this period, were relatively richer. (For example, the owner of the maximum number in 2001/2 keeping 10000 poultry and 4 buffaloes was also the maximum keeper five years earlier, holding 5000 poultry and 4 buffaloes. So this farm doubled the poultry holding, and it belonged to the richest quintile. Proportionately more of other relatively larger farms, no matter whether increasing or decreasing their size of holding, belonged to the relatively richer income quintiles, except for those owning goat/sheep). Examining the households having more than 10 heads, we observed that about 44% of the goat rearers in the poorest quintile belonged to this group in comparison to 41% of the goat rearers in the richest quintile. For sheep, the proportion was 65% and 60% respectively. Average sheep holding for the sheep keeping households were double the average goat holding of the goat keeping households. So for sheep, we examined the holding above 15 and found that the proportions for the poorest and the richest quintiles were 45% and 48% respectively. This was due to 3 large sheep holders in the richest group with 40 or more animals. The poorest quintile had a maximum size of 30 and this suggests that similar to poultry, poorer sheep keepers failed to increase the size permitted by the available technology.  

Moving on the panel households re-interviewed in the early 2005 for the 2003/4 stocks, we discovered a different trend to that observed in between 1996/7 and 2001/2. Over these more recent three years both farming density as well as herd size decreased; exceptions were calves, small ruminants and chicken (Table 5).

A decreasing trend in the farming density, after an increasing trend in the previous period, seems to be mostly due to the bad drought over the three year period. An increase in calf farming and at the same time a decline in buffalo and cow farming indicate that the farms without the possibility of having new calves discontinued farming. This indicates a shift of technology towards stall feeding intensive milk production by fewer farms with improved milch bovines having the capability of giving birth as developed in some of the villages in response to the growing scarcity of open grazing. These milch bovines may be able to produce higher quantity of milk. This is supported by the evidence that out of 82 buffalo farms in 2003/4, 75 farms were selling milk/curd and all of the cow farms in 2003/4 were selling milk/curd.  The decrease in bullock farming may be attributable inter alia to tractorisation to a large extent. 

For chickens, both farming density and flock size increased in the 1996/7 to 2001/2 period but in the more recent period farming density declined with the increase in the average flock size at a higher rate. The increase in flock size on an average was due to a few large poultry farms indicated by the increase in maximum size from 1000 birds to 2000 birds. As discussed earlier, the large farms were mostly wealthier households implying that the better offs benefited more from the change in technology.

In the panel sample, the average of total herd size was higher in 2001/2 than in 1996/7, unlike the census. For this sample, the growth of farming was lower than in the census. During the 2001/2 to 2003/4 period, the average herd size grew at a faster rate but the farming density declined. The faster growth in herd size was due to intensive poultry production by a few large farms. The overall change was not statistically significant at the 5% level but in the sample without 3 intensive poultry farms a significant negative trend was found for this period of 2001/2 to 2003/4, corresponding with the census data.

The decline in livestock population was spread over a large number of households (52.6% in Table 6).

A large proportion of households reported an increase in the number of calves and goats. The increase in farms having calves supports our earlier argument on technical change and these farms were located mostly in the villages of Chittoor district where the privatization of commons as well as long standing drought badly affected the access to CPRs. Households that reported increased goat herds were relatively poorer and mostly located in the poorer villages such as KA and MD. These households might acquire goats as a coping strategy in the face of long standing drought.
 

Reasons associated with the change

Of the farms which managed to increase livestock, 56% appeared to get access to better markets (Table 7). 

The next important reason was the access to more labor input followed by access to grazing/fodder. About 7% of the farms reported that they increased their livestock numbers by using the money from migration and remittances. These farms were either marginal or small, were from the poorest 3 quintiles, and from backward castes. This indicates that the use of migration income to invest in livestock enterprise was concentrated in the poorer groups.

Domestic shocks or stresses were identified as the most important cause for a decrease in numbers of livestock, followed by pest and disease problems. Other natural/environmental factors such as drought, and loss of CPRs were also identified as important reasons. The loss of access to grazing/fodder resulted both from natural factors like the drought as well as man made factors like CPR related rules and regulations such as the privatization of commons and overgrazing. The literature reported that the restrictions on the use of common grazing areas under the watershed projects in Andhra Pradesh had forced landless livestock owners to sell their stock but landowners benefited by producing more crop residues using improved irrigation facilities due to the project (Turton 2000). Natural/environmental factors altogether were identified a major cause of decrease in the livestock population.

About 37% of the farms which experienced a decline in stock, reported the selling of livestock due to domestic shocks or stresses. This involved 39% of the total large ruminant farms and 55% of the total small ruminant farms. Thus, although fewer farms were keeping small ruminants, in terms of buffering function their proportionate contribution was large when we consider the households who keep them.

We combined these factors associated with the decrease in livestock into three major groups to carry out an analysis based on income quintiles to find out the situation of the poorer households (Table 8).

All quintiles apart from the richest, identified shock/stress variables as the most important, the richest identifying the poor natural environment as the most important.
 

Impact of the change in income of the poor

Among the households selected for the panel survey in 2005, a sample of 145 had reported income from livestock in 2001/2 census survey and a sample of 208 reported income from livestock (Table 9) (91 households were common in these samples, others reported livestock income either in 2001/2 or in 2003/4).

These households earned about 27% income share in 2001/2 and 25% income share in 2003/4. Thus the livestock share in household income had fallen in three years period but the share in the poorest group had risen indicating an increase in dependency of the poorer on livestock.

The proportion of gross income earned by the poorest quintile of a sample of livestock growers was found as high as 65% in 2003/4 and 54% in 2001/2. This sample consists of households reporting cash income from livestock. Out of 208 farmers in 2003/4, 15 in the poorest quintile were earning all of their income from livestock.  Ten households in the poorest quintile reported regular income from livestock in both periods and their average share was as high as 80%. Some of these households are expected to be mobile from one quintile to another in this three year period, but this is not taken into account in this analysis due to absence of total income data for 2003/4. Taking the sample of all livestock keepers, either in 2001/2 or in 2003/4 or in both periods, we find that the average share of livestock income was 17% in 2001/2, increasing to 22% in 2003/4 (This calculation is based on the total income data collected in 2001/2 adjusted for inflation and given the increasing growth situation in India it is possible that the share in 2003/4 would be lower than this calculated figure. The large average share of the poorest group indicates that the share would still be higher than the previous period even if we could consider the increased income in 2003/4). The average share of the households reporting regular income in both periods was almost the same in 2001/2 and 2003/4 and it was about 30% but still the average share was much lower in 2001/2 than in 2003/4 5 Out of 236 households keeping livestock in any or both the periods, 91 households had direct income from this source in both periods, 145 reported income in 2001/2 and 208 reported income in 2003/4 (Out of 236 households keeping livestock in any or both the periods, 91 households had direct income from this source in both periods, 145 reported income in 2001/2 and 208 reported income in 2003/4).

Conclusions and implications

• This paper investigates the factors associated with changes in the species composition in smallholder livestock keeping and how these changes are affecting the income of the poor in the Indian State of Andhra Pradesh. The data used for the study are drawn from Livelihood Options Study of Overseas Development Institute, London, specifically the Census Survey 2001/2 and Panel Survey 2003/4. About 40% of the census households were keeping one or more types of livestock species in 2001/2. Of the panel households 52% were raising livestock in 2001/2 but this fell by two percentage points by 2003/4.

  
  

• The variables most strongly associated with livestock ownership were land, diversification and species. Assets, landlessness and migration were also important determinants. The richer households responded relatively more with respect to land. Poorer households which had at least one migrating member had much less ownership than the richer households with migration. The impact of livelihood diversification on livestock ownership was higher for the poorer households. Households owning single species had lower numbers of livestock, suggesting that few had succeeded in specialising, and that the majority pursued risk aversion strategies of keeping several different types of livestock. Diversification of species seems to be the result of some adaptive mechanism to cope with income risk under the changing environment. Poorer households depend disproportionally on livestock and on common pool resources as a means of feeding them. These resources are becoming scarcer and prompt environmental policies which in turn make it more difficult for the poor to maintain preferred livestock types. Most of the farms, in particular the poorer groups diversified their holdings by decreasing the numbers kept of a single species. The average herd size of any single species declined considerably in the five year period beginning from 1996/7, but the number of farms keeping the species rose, indicating an increase in species diversification. The trade off between the ownership of a species and diversification failed to increase the total herd size, the drop in one species was more than offset by the rise in other species within a farm. Over the more recent three years of 2001/2 to 2003/4 both farming density as well as herd size had decreased. A decreasing trend in the farming density, after an increasing trend in the previous period, seems to be mostly due to the drought experienced over the three year period. Species was also identified as an important determining factor for livestock ownership in a multivariate analysis. About 32% of the livestock keepers were landless and 70% of them kept more than one species. The poorer households kept fewer small ruminants in the latter period but richer households kept more. An increase in calf farming and at the same time a decline in buffalo and cow farming indicate that the farms without the possibility of having new calves discontinued farming. The decrease in bullock farming may be attributable inter alia to more tractorisation. These all are indicative of the shifting of technology towards what the better off can afford. Policies such as the ban on goats will also have affected the poor more since they are more visible – i.e. using open-grazing methods.

  
  

• Proximity to townships reduces livestock keeping as because of more factor market linkages, less access to common pool resources, better infrastructure and institutional facilities. The village which is far but has access to these facilities kept fewer livestock, township characteristics cannot be a single identifier of livestock supply.  

  
  

• The proportion of gross income earned by the poorest quintile of a sample of livestock growers was found as high as 65% in 2003/4 and 54% in 2001/2. Out of 208 farmers in 2003/4, 15 in the poorest quintile were earning all of their income from livestock. 

  
  

• Most farmers identified shock/stress variables as most important for the decrease in livestock population. Better management of common pool resources is urgently required to make more fodder resources available through cut and carry systems.  Reasons like paying of debt, purchase of inputs, medical treatment and marriage expenses are contributing to distress sales and so need to be addressed through better institutional arrangements. More than 22% of farmers mentioned disease problems as a cause of the decline in livestock population. This problem needs to be addressed, possibly through public-private partnership arrangements as are being tried by several organizations in different countries.

  
  

• Domestic shocks or stresses were identified as the most important cause for a decrease in numbers of livestock, followed by pest and disease problems. Other natural/environmental factors such as drought, and loss of CPRs were also identified as important reasons. The loss of access to grazing/fodder has resulted both from natural factors like the drought as well as man made factors like CPR-related rules and regulations such as privatization of commons and overgrazing. The literature reported that the restrictions on the use of common grazing areas under the watershed projects in Andhra Pradesh had forced landless livestock owners to sell their stock but landowners benefited by producing more crop residues using improved irrigation facilities due to the project.

References

De Haan C, van Veen T C, Brandenburg B, Gauthier J, Le Gall F, Mearns R, and Simeon M. 2001 Livestock Development: Implications for Rural Poverty, the Environment, and Global Food Security, The World Bank, Washington, D.C. pp 96. http://www-wds.worldbank.org/servlet/WDSContentServer/WDSP/IB/2001/12/11/000094946_01112104010387/Rendered/PDF/multi0page.pdf

Deshingkar P, Farrington J, Akter S, Rao L, Sharma P, Freeman A, and Reddy J. 2006 Livestock and Poverty reduction in India, International Livestock Research Institute and Overseas Development Institute, Draft Working Paper.

FAO 2005 Evaluation of Livestock Production, Policy and Information (Programme 2.1.3), Report of the External Peer Review Panel, FAO, Rome. http://www.fao.org/docrep/meeting/009/j4779e/j4779e00.htm

Farrington J,  Deshingkar P, Johnson C and  Start D (ed) 2006 Policy Windows and #Livestock Research for Rural Development 19 (11) 2007Livelihood Futures: Prospects for Poverty Reduction in Rural India, Oxford and New Delhi, OUP.

Farrington J and Lobo C 1997 Scaling up participatory watershed development in India: lessons from The Indo-German Watershed Development Programme. Natural Resource Perspectives 17. pp 6. http://www.odi.org.uk/NRP/17.html

Holmann F, Rivas L, Urbina N, Rivera B, Giraldo L A, Guzman S, Martinez M, Medina A and Ramirez G 2005 The role of livestock in poverty alleviation: An analysis of Colombia. Livestock Research for Rural Development 17 (1). http://www.cipav.org.co/lrrd//lrrd17/1/holm17011.htm

Turton C. 2000 Enhancing livelihoods through participatory watershed development in India, Working Paper 131, Overseas Development Institute, London. pp 28. http://www.odi.org.uk/publications/wp131.pdf

Received 19 July 2007; Accepted 5 September 2007; Published 1 November 2007

Go to top