# Integrated bio-economic simulation model for sheep production: Ex-ante evaluation of investment opportunities in Ethiopia

←

**Page content transcription**

If your browser does not render page correctly, please read the page content below

LIVES WORKING PAPER 9 Integrated bio-economic simulation model for sheep production: Ex-ante evaluation of investment opportunities in Ethiopia ISBN 92–9146–436–8 Livestock and irrigation value chains for Ethiopian smallholders project aims to improve the competitiveness, sustainability and equity of value chains for selected high‐value livestock and irrigated crop commodities in target areas of four regions of Ethiopia. It identifies, targets and promotes improved technologies and innovations to develop high value livestock and irrigated crop value chains; it improves the capacities of value chain actors; it improves the use of knowledge at different levels; it generates knowledge through Livestock and Irrigation Value chains for Ethiopian Smallholders action‐oriented research; and it promotes and disseminates good practices. Project carried out with the financial support of the Government of Canada provided through Foreign Affairs, Trade and Development Canada (DFATD). lives-ethiopia.org The International Livestock Research Institute (ILRI) works to improve food security and reduce poverty in developing countries through research for better and more sustainable use of livestock. ILRI is a member of the CGIAR Consortium, a global research partnership of 15 centres working with many partners for a food-secure future. ILRI has two main campuses in East Africa and other hubs in East, West and southern Africa and South, Southeast and East Asia. ilri.org The International Water Management Institute (IWMI) is a non-profit, scientific research organization focusing on the sustainable use of water and land resources in developing countries. It is headquartered in Colombo, Sri Lanka, with regional offices across Asia and Africa. IWMI works in partnership with governments, civil society and the private sector to develop scalable agricultural water management solutions that have a real impact on poverty reduction, food security and ecosystem health. IWMI is a member of CGIAR, a global research partnership for a food-secure future. iwmi.org CGIAR is a global agricultural research partnership for a food-secure future. Its science is carried out by15 research centres that are members of the CGIAR Consortium in collaboration with hundreds of partner organizations. cgiar.org EIAR

Integrated bio-economic simulation model for sheep production: Ex-ante evaluation of investment opportunities in Ethiopia i Integrated bio-economic simulation model for sheep production: Ex-ante evaluation of investment opportunities in Ethiopia Asfaw Negassa1, Berhanu Gebremedhin1, Solomon Desta1, Kidus Nigussie2, Getachew Gebru1, Barry Shapiro1, Celine Dutilly-Diane3 and Azage Tegegne1 1. International Livestock Research Institute 2. Ministry of Agriculture, Ethiopia 3. French Agricultural Research Centre for International Development (CIRAD) October 2015

ii Integrated bio-economic simulation model for sheep production: Ex-ante evaluation of investment opportunities in Ethiopia © 2015 International Livestock Research Institute (ILRI) This publication is copyrighted by the International Livestock Research Institute (ILRI). It is licensed for use under the Creative Commons Attribution-Noncommercial-Share Alike 3.0 Unported Licence. To view this licence, visit http:// creativecommons.org/licenses/by-nc-sa/3.0/. Unless otherwise noted, you are free to copy, duplicate or reproduce, and distribute, display, or transmit any part of this publication or portions thereof without permission, and to make translations, adaptations, or other derivative works under the following conditions: ATTRIBUTION. The work must be attributed, but not in any way that suggests endorsement by ILRI or the author(s). NON-COMMERCIAL. This work may not be used for commercial purposes. SHARE ALIKE. If this work is altered, transformed, or built upon, the resulting work must be distributed only under the same or similar licence to this one. NOTICE: For any reuse or distribution, the licence terms of this work must be made clear to others. Any of the above conditions can be waived if permission is obtained from the copyright holder. Nothing in this licence impairs or restricts the author’s moral rights. Fair dealing and other rights are in no way affected by the above. The parts used must not misrepresent the meaning of the publication. ILRI would appreciate being sent a copy of any materials in which text, photos etc. have been used. Editing, design and layout—ILRI Editorial and Publishing Services, Addis Ababa, Ethiopia. Cover photo—ILRI/Zerihun Sewunet ISBN: 92–9146–436–8 Citation: Negassa, A., Gebremedhin. B., Desta, S., Nigussie, K., Gebru, G., Shapiro, B., Dutilly-Diane, C. and Tegegne, A. 2015. Integrated bio-economic simulation model for sheep production: Ex-ante evaluation of investment opportunities in Ethiopia. LIVES Working Paper 9. Nairobi, Kenya: International Livestock Research Institute (ILRI). ilri.org better lives through livestock ILRI is a member of the CGIAR Consortium Box 30709, Nairobi 00100, Kenya Box 5689, Addis Ababa, Ethiopia Phone: + 254 20 422 3000 Phone: +251 11 617 2000 Fax: +254 20 422 3001 Fax: +251 11 617 2001 Email: ILRI-Kenya@cgiar.org Email: ILRI-Ethiopia@cgiar.org

Integrated bio-economic simulation model for sheep production: Ex-ante evaluation of investment opportunities in Ethiopia iii Contents Tables iv Figures vi 1 Introduction 1 1.1 Background and justification 1 1.2 Objectives of the study 2 2. The data and the scope of the study 3 3. Analytical framework 4 4. Empirical approach 5 4.1 Stage-structured sheep flock projection matrix model 5 4.2. Economic model 9 5. Results and discussions 12 5.1 Results of baseline assessments 12 5.2 Ex-ante impact assessment of policy and investment interventions 25 6. Conclusions and policy recommendations 29 7. References 31

iv Integrated bio-economic simulation model for sheep production: Ex-ante evaluation of investment opportunities in Ethiopia Tables Table 1. Baseline initial sheep flock size and structure for agro-pastoral and pastoral sheep production systems 12 Table 2. Initial stage specific demographic rates used in calibrating sheep flock projection matrices for agro-pastoral and pastoral sheep production systems 13 Table 3. Annual Lefkovitch stage-structured population projection matrices for sheep flocks in agro-pastoral and pastoral sheep production systems 13 Table 4. Baseline stable stage distribution (w), stage-specific reproductive values (v) and annual long-term population multiplication rate for agro-pastoral and pastoral sheep production systems 14 Table 5. Sensitivity and elasticity matrices from Lefkovitch stage-structured population projection matrix for agro-pastoral and pastoral sheep production systems 16 Table 6. Steady-state stable stage distribution (w), stage specific reproductive values (v) and offtake rates for agro-pastoral and pastoral sheep production systems 16 Table 7. Steady state sheep meat production and annualized present value of gross margins for agro-pastoral and pastoral sheep production systems in Ethiopia 17 Table 8. Baseline initial sheep flock size and structure for the MRD sheep production system 17 Table 9. Stage specific demographic rates used in calibrating sheep flock projection matrices for the MRD production system 18 Table 10. Lefkovitch sex-structured population projection matrices for the MRD sheep flock 18 Table 11. Baseline stable stage distribution (w), stage specific reproductive values (v) and annual long-term population multiplication rate for the MRD sheep production system 19 Table 12. Sensitivity and elasticity matrices from Lefkovitch stage-structured population projection matrix for the MRD sheep production system 20 Table 13. Steady-state stable stage distribution (w), stage specific reproductive values (v) and offtake rates for the MRD sheep production system 20 Table 14. Steady state sheep meat production and annualized present value of gross margins by different sheep production systems in Ethiopia 21 Table 15. Baseline initial sheep flock size and structure for the MRS sheep production system 21 Table 16. Stage specific demographic rates used in calibrating sheep flock projection matrices for the MRS sheep production system 22

Integrated bio-economic simulation model for sheep production: Ex-ante evaluation of investment opportunities in Ethiopia v Table 17. Annual Lefkovitch stage-structured population projection matrices for the sheep flock MRS production system 22 Table 18. Baseline stable stage distribution (w), stage specific reproductive values (v) and annual long-term population multiplication rate for the MRS sheep production system 23 Table 19. Sensitivity and elasticity matrices from Lefkovitch stage-structured population projection matrix for the MRS sheep production system 23 Table 20. Steady-state stable stage distribution (w), stage specific reproductive values (v) and offtake rates for the MRS sheep production system 24 Table 21. Steady state sheep meat production and annualized present value of gross margin for the MRS sheep production system in Ethiopia 24 Table 22 Assumptions on annual recurrent costs associated with investment costs to reduce young stock mortality (ETB/head) 26 Table 23. Projected national production, consumption, and production and consumption balance of sheep meat ‘with’ and ‘without’ combined investment interventions in Ethiopia by year 2028 27 Table 24. Returns to combined policy and investment interventions in cattle production by sheep production systems in Ethiopia (2013–32) 28 Table 25. Summary of discounted benefits and costs due to combined investment interventions in sheep production by production system in Ethiopia 28

vi Integrated bio-economic simulation model for sheep production: Ex-ante evaluation of investment opportunities in Ethiopia Figures Figure 1: Number of sheep by sex and stage of growth; Agro-pastoral herd 14 Figure 2: Number of sheep by sex and stage of growth, Pastoral small herd 15 Figure 3: Number of sheep by sex and stage of growth, Pastoral medium herd 15 Figure 4: Number of sheep by sex and stage of growth, Mixed moisture deficient small herd 19 Figure 5: Number of sheep by sex and stage of growth, Mixed moisture deficient medium herd 19 Figure 6: Number of sheep by sex and stage of growth, Mixed moisture surplus small herd 23

Integrated bio-economic simulation model for sheep production: Ex-ante evaluation of investment opportunities in Ethiopia 1 1 Introduction 1.1 Background and justification Ethiopia’s livestock production is characterized by mixed-species and multi-purpose type and the development of intensive commercially oriented specialized (single species or single purpose) livestock production is at its early stage. Sheep are found to be an integral part of these types of livestock production systems. There are about 15 known major local sheep breeds raised in different agro-ecological zones of the country (Awgichew and Abegaz 2008). Ethiopia has the largest sheep population in Africa. In 2013, the most recent comprehensive estimate of sheep population, based on the compilations of data from different sources, indicate that there are about 29 million heads of sheep distributed among the major livestock production systems in the country; agro-pastoral and pastoral (50%) and the highland mixed crop–livestock production systems (50%.) (LSA, forthcoming 2016). In general, sheep are raised in Ethiopia both in the highland mixed crop–livestock production systems and in the lowland agro-pastoral and pastoral livestock production systems. Similar to other ruminants like cattle and goats, sheep also provide multiple outputs to livestock keepers; they are a source of relatively affordable nutritious food (meat); produce organic manure for crop production; generate cash income from the sale of live sheep, sheep meat or skins; and represent a form of asset that can easily be liquidated for emergency situations and small cash requirements. In Ethiopia, dairy sheep or keeping sheep for milk production is not known. In some areas, sheep are also raised for wool production. In general, the potential of sheep contribution towards improving farm households’ income and food security situations and national economic growth in developing countries like Ethiopia is tremendous. However, despite its economic potential, sheep production is also besieged by several constraints affecting its productivity and production and hence limiting its contribution towards improving farm household income and food security, poverty reduction and national economic growth. Similar to goats production system, the sheep production system in Ethiopia is also characterized by poor genetic performance and a high level of morality due to poor management practices, feed scarcity, and inadequate animal health services and prevalence of animal diseases (Gebremariam et al. 2013; Gizaw et al. 2013; Haile et al. 2011; Negassa et al. 2011; Gizaw et al. 2010; Negassa and Jabbar 2008; Tolera 2007). Sheep are kept by large number of pastoral, agro-pastoral and smallholder farmers in Ethiopia. For example, Negassa et al. (2011) indicate that 20% and 38% of the agro-pastoral and pastoral households and smallholder farmers, respectively, own sheep. However, the size of their ownership is very small and their productivity level is also observed to be one of the lowest in eastern Africa (Negassa et al. 2011). For sheep producers to benefit from the rising demand for sheep meat and products, they need to have an adequate number of sheep that can generate sufficient marketable surpluses of sheep on sustainable basis. Particularly, given the limited land resources for feed production and environmental constraints to increasing livestock production, it is important to increase the productivity of the sheep production from the limited existing sheep flock and land holdings. In this regard, an understanding of the baseline sheep performance and flock growth dynamics is very important to examine the evolution of sheep flock size and structure over time and to identify potential targets for different interventions to increase sheep producers’ profits and their volume of participation in live sheep and sheep products markets on sustainable basis.

2 Integrated bio-economic simulation model for sheep production: Ex-ante evaluation of investment opportunities in Ethiopia The full realization of the potential of sheep require investment to improve sheep production management, especially in feeding and disease control for improving sheep growth and reproductive rates along with creating an enabling environment (institutions, infrastructure and policies) to allow producers to take advantage of growing sheep market opportunities. However, the investment resources are very scarce and detailed sheep sub-sector sector analyses in terms of identifying and prioritizing investment opportunities and constraints in the sub-sector sector are critical for effective and efficient utilization of the scarce investment resources. Currently, there are no comprehensive ex-ante impact assessments of policy and investment interventions and ex-ante analytical and empirical modelling frameworks for guiding and prioritizing public and private investment decisions to promote the development of sheep sub-sector in Ethiopia. This paper develops and implements an empirical bio-economic sheep flock growth model to analyse the sheep’s flock growth dynamics and its economics to inform the various policy changes and investment strategies to develop the sheep sub-sector. It builds on previous detailed livestock sector analyses for Ethiopia by the International Livestock Research Institute (ILRI) jointly with the Livestock State Ministry of Ethiopia using the Livestock Sector Investment and Policy Toolkit (LSA) developed by the World Bank and Centre de Coopération Internationale en Recherche Agronomique pour le Développement (CIRAD) (LSA 2013). It focuses on detailed analyses of sheep flock growth dynamics for representative livestock production systems which allows ex-ante impact assessment of policy and investment interventions and thus can inform and guide strategic development in infrastructure, institutional, technological and management practices that support significant increase in sheep production, productivity and commercial offtake. The modelling framework provides information on economics of sheep production at different scales (flock level and production system level) and can be aggregated at regional or national level, thus making it possible to prioritize investment in the research and development of sheep at regional or national level. 1.2 Objectives of the study The overall objective of this paper is to develop and apply an analytical and empirical modelling framework which integrates a sheep flock growth model with and economic model for simulating sheep meat and milk production for ex-ante assessment of the financial profitability of investment interventions to improve sheep production in mixed crop–livestock and agro-pastoral and pastoral production systems in Ethiopia. The specific objectives are: 1. To provide a baseline characterization of sheep flock demographics and flock growth dynamics over time in terms of the sheep flock growth rate, offtake rate, size and structure for representative production systems and flock size classes; and 2. To evaluate ex-ante the technical and financial impacts of combined investment interventions in sheep health and feeds management on sheep production. This paper is organized as follows. The next section provides brief description of the data used in the analysis. Section 3 presents analytical framework guiding the empirical analysis. The empirical model used is presented in section 4, while section 5 presents the results and discussions. Finally, the conclusions, policy implications and recommendations are made in section 6.

Integrated bio-economic simulation model for sheep production: Ex-ante evaluation of investment opportunities in Ethiopia 3 2. The data and the scope of the study The detailed discussion of the data sources used in this analysis is documented in the livestock sector analysis report for Ethiopia (LSA forthcoming 2016). The data has been obtained from different secondary sources. First, the initial sheep flock size and structure data is generated from the Living Standard Measurement Survey (LSMS) data conducted in 2010/11 jointly by the World Bank and the Central Statistical Authority (CSA) of Ethiopia (LSMS 2010/11) and from Save the Children UK livestock survey data for pastoral areas of Ethiopia. Second, the sheep demographic data (for example, fertility rates, mortality rates and offtake rates) used to generate parameters for various sheep flock projection matrices was collected using the Delphi technique which involved the use of a structured questionnaire to collect technical data from the national and international livestock experts. The data collected using the Delphi method was validated through panels of expert meetings and discussions. Third, the costs of production for different livestock enterprises and prices for the valuation of livestock and livestock products were also obtained from the expert interviews and secondary sources. This study focuses on agro-pastoral and pastoral and mixed crop livestock production systems which account for most sheep population and livestock dependent population. There is a trend towards intensive specialized commercial livestock production systems, but the scale is still not significant, and therefore, it is not considered in this paper.

4 Integrated bio-economic simulation model for sheep production: Ex-ante evaluation of investment opportunities in Ethiopia 3. Analytical framework The analytical framework for Livestock Sector Investment Policy Toolkit (LSIPT) is a dynamic bio-economic simulation modelling. A non-stochastic dynamic sheep flock growth model is used for the projection of the sheep population and outputs from sheep production (meat, or manure) over a fifteen year time period. The model also allows the simulation of sheep production outputs given inputs required for sheep production. Then, the flock model is linked with an economic model to allow the evaluation of financial and economic profitability of sheep production over the same time period based on a proper valuation of goat production inputs and outputs. The simulation analysis allows the assessment of the technical and financial performance of the sub-sector over time given the input use, existing technologies, market conditions and policies facing the sub-sector. The analysis is conducted at flock level, but the results of the simulation analysis can be aggregated at different levels: production system level, at regional and national level. The baseline assessment provides a counterfactual scenario (‘without’ intervention) against which new sheep policy and investment interventions can be evaluated. Thus, it provides the basis for ex-ante impact assessments and benefit–cost analyses of various proposed interventions aimed at enhancing the contribution of the sheep sub-sector to farm household income growth, food security, poverty alleviation and national economic growth.

Integrated bio-economic simulation model for sheep production: Ex-ante evaluation of investment opportunities in Ethiopia 5 4. Empirical approach 4.1 Stage-structured sheep flock projection matrix model The population projection matrices are based on Leslie (1945); Caswell (1989); and Caswell (2001). Recent applications of population projection matrices to the study of livestock population growth dynamics include: Lesnoff (1999); Lesnoff (2000); Lesnoff et al. (2009); and Lesnoff et al. (2012). In general, there are two types of population projection matrices which are widely used in analysing livestock and plant population growth dynamics over time. The first matrix model is called the Leslie population projection matrix model which classifies the livestock population by age, hence an age-structured matrix model (Leslie 1945). The second matrix model is called the Lefkovitch population projection matrix model which classifies livestock population by their stage of growth or life cycle and hence is called the stage-structured population projection matrix model (Lefkovitch 1965). Each approach has its advantages and disadvantages. The Leslie matrix is considered to be data intensive (for example, requires age-specific demographic data) and mathematically more demanding (due to higher dimensions of matrices used) as compared to the stage-structured Lefkovitch matrix model. Moreover, in terms of application, the stage- structured matrix is more practical, for example, as the utilizations of sheep or any livestock for different purposes are mainly based on stages of growth (age ranges) rather than age-specific per se. Both matrix models project the population in t+1 time period using the initial population at t time period and annual transition parameters. For both population projection matrices, the availability of reliable demographic data (e.g. fecundity rates and mortality rates) is critical for the accuracy of prediction. In this paper, the Lefkovitch sex and stage-structured population projection matrix model is used due to its relative simplicity computationally and realistic characterization of sheep growth stages. In a Lefkovitch stage-structured sheep population projection matrix model, the structure of the sheep flock is defined in terms of animal numbers in different age classes and sex cohorts. Thus, the female and male sheep are divided in to three discrete growth stages based on age: juvenile, sub-adult and adult. In general, the stage durations for sub- adults and adult sheep are assumed to be different by sex of sheep and by production system analysed. The projection interval, or time step, used for the projection is one year and the overall sheep population projection time horizon is 20 years. The flock growth model is run by different flock size classes and for major production systems to see the effects of flock size and production system on sheep flock growth dynamics. Mathematically, the stage-structured deterministic (non-stochastic) population projection matrix model for the analysis of sheep flock growth dynamics considered here is given as a discrete time first-order difference equation: (1) Where nt+1 denotes the 6x1 state vector of sheep flock sizes by sex and by stage of sheep growth at a time t+1; nt is a 6x1 state vector of sheep flock size by sex and stage of sheep growth at time t; and A is a square 6x6 Lefkovitch annual stage-structured population projection matrix used to analyse sheep population dynamics in discrete time (Lefkovitch 1965). The A matrix is used to generate a new state variable or vector nt+1 and contains the annual transition probabilities derived based on annual demographic rates (annual fecundity rates, annual mortality rates and annual offtake rates). In general, the Lefkovitch population projection matrix A is a generalization of the Leslie age- structured matrix and is given as:

6 Integrated bio-economic simulation model for sheep production: Ex-ante evaluation of investment opportunities in Ethiopia (2) Where Fa is the fecundity of adult female sheep which is the product of annual parturition rate and net prolificacy rate, the subscripts j, s and a denote juvenile, sub-adult and adult goat, respectively irrespective of the sex of the sheep and Gi,j is the probability that a sheep of a given sex in stage i will enter the next stage j (j=i+1) in the next time period; and Pi,i denotes the probability of an individual sheep of a given sex surviving and remaining (or persisting) in the same stage i. The structure and formulation of Leslie and Lefkovitch matrices are similar but the difference is in terms of the columns and matrix entries. In the case of Leslie matrix, the columns of matrix A represent the age of the sheep or an animal, while in the case of Lefkovitch matrix the column represents the different stages of growth. The other difference between the two is that the matrix entries in the case of the Leslie matrix is given in terms of fecundities in the first row which indicate the reproductive contribution for different sheep growth stages and survival probabilities (pi) across the diagonals of the matrix. On the other hand, in the case of Lefkovitch matrix, the fecundities are given in the first row, but the survival probabilities are broken down into two: the probability of an individual surviving and moving from class i to the stage j (Gi,j) and the probability of an individual surviving and remaining (or persisting) in the same stage i (Pi,i). Given the duration of sheep in each stage (di) and the stage-specific survival probabilities, the matrix A entries for Gi and Pi are computed using the following formulas (Crouse et al. 1987): (3) (4) where pi is the annual survival rate computed below following Lesnoff et al. (2000) based on sex and stage specific demographic rates: annual mortality rate (mi) and sex-stage specific annual offtake rate (oi). The survival probability is given as follows: (5) Equation (5) indicates a self-recruiting flock growth model or endogenous population dynamics model where animals are not imported from outside the population to accelerate growth (Lesnoff 2012). Thus, the sheep demographic behaviour is assumed to be influenced mainly by both mortality and commercial offtake rates only. Finally, the first order difference equation (1) can be given in vectors and matrix representations as follows: * (6) Where the subscripts in the A matrix denote the stages of sheep growth in i and i+1 and the multiplication of the fecundity rate by 0.5 is to indicate the 1:1 female to male ratio at birth assumed in the model since this represents a natural breeding system. Note that the matrix A is divided into sex-stage specific blocks as indicated by the horizontal line, the entries in the upper block are for the female sheep (F), while the entries in the lower blocks are for the male sheep (M). The basic assumption of the above transition matrix is that only adult females can produce newborn calves hence female dominant matrix model (Lesnoff 1999).

Integrated bio-economic simulation model for sheep production: Ex-ante evaluation of investment opportunities in Ethiopia 7 Once the transition matrix A is setup, the next step in the analysis of sheep flock dynamics is to find out the eigenvalues and eigenvectors of the population projection matrix A. Several important parameters which characterize the sheep flock dynamics emerge from the eigenvalue and eigenvector analyses of the transition matrix A. First, the dominant eigenvalue (λ) of matrix A represents the finite (asymptotic) sheep population multiplication rate, while log (λ) gives intrinsic annual growth rate which is the continuous growth rate per individual sheep in the population. Thus, in the long-run the sheep population follows exponential growth rate given by log (λ) or: (7) (8) Since A is 6X6 square matrix there are six possible eigenvalues and six associated eigenvectors with matrix A. However, the annual rate of increase of population is given by the dominant eigenvalue. The eigenvalues are defined as the solutions to the characteristic equation: (9) Where det denotes determinant and I is an identity matrix. The sign of λ indicates whether the flock size (or population) is declining (λ 1) in the long run and shows that potential exists for increased commercial offtake above the current level. For example, λ equal to 1.2 means population increases by 20% per year and λ equal to 0.93 means the population will decrease by 7% per year over the long run. Second, the normalized eigenvector (w) associated with the dominant eigenvalue gives the stable stage-structured proportion for sheep flock size. Third, the standardized eigenvectors (v) associated with the dominant eigenvalue of the transpose of matrix A provide the reproductive values of different stages of sheep production as compared to the juvenile, this measures the relative contributions of different stages of sheep growth to long-term growth in sheep flock size. It is also important to note that several important simplifications of the presentation of the projection matrix model can be made once the dominant eigenvalue and the associated eigenvector are determined. These simplifications are useful for projection purposes. First, there is equivalence between the transition matrix and the dominant eigenvalue and as a result the following relationship holds at equilbrium: (10) From equation (10) it follows that the sheep projection equation (1) can be alternatively given as: (11) For T large, λ is also approximated by annual empirical multiplication rate given as the ratio of Nt+1 to Nt where N represents total herd size for all age groups and sex at a given time. Equation (1) can be also generalized to give the sheep projection model at any time t given the initial flock size and stable flock proportion as: (12) 0 Where nt is a vector of sheep population at time t; n0 is scalar and denotes the initial sheep flock size; λ is the dominant eigenvalue and w is the normalized eigenvector associated with the dominant eigenvalue and denotes stable stage population proportions. Similarly, given the stable sex and stage distribution, and asymptotic sheep growth rate, the total commercial live sheep offtake for sheep with ith sex in jth growth stage at any time t can be obtained using the following commercial offtake function: (13)

8 Integrated bio-economic simulation model for sheep production: Ex-ante evaluation of investment opportunities in Ethiopia Where Oij,t is a vector of commercial offtake of the ith sex in jth growth stage at t time. The total carcass equivalent commercial live sheep offtake at a given time period is given by aggregating the carcass offtake rates across sex and stage classes. This shows the quantity of commercial offtake with constant sheep growth rate and stable stage- structure w. It is also possible to compute steady-state commercial offtake which is characterized by equilibrium situation whereby the growth rate is zero. The simulation analysis to see the effects of different interventions is usually conducted for steady state situation to avoid bias due to change in sheep flock size and structure. For steady-state situation since λ is equal to 1, the sheep flock projection model given in equation (12) is modified as: (14) and accordingly the commercial offtake function can be also modified as: (15) The steady state offtake rate is a constant while the sheep flock size or population remains constant over time. This is a desirable management strategy whereby a regular offtake is made thereby producing regular income and returning sheep population to stable size and structure. This is important particularly when there is a need to limit the size of the sheep population due to the need to match livestock numbers to available feed resources. This objective is achieved through numerical manipulation of the transition matrix whereby the λ is equated to 1 and the offtake rates are choice variables selected using non-linear optimization technique. One of the important goals of the analysis of sheep flock growth dynamics is to determine what stages and what demographic rates are most likely causing the increase or decrease in the sheep flock. This is important because resources are limited so priority areas of development and research interventions need to be identified. For this purpose, sensitivity and elasticity analyses are made in order to assess the impacts of changes in sheep demographic factors (reproductive rates and mortality rates) on long-term sheep growth rate (λ). The sensitivity is calculated from the elements of the right and left eigenvectors (w and v, respectively) as follows (de Kroon et al. 2000): (16) Where v is the left eigenvector associated with A, and is the scalar product of vectors w and v. The sensitivity analysis estimates the impact of an absolute change in the demographic rates on population growth rate. High sensitivity values indicate that a small intervention could have a large impact in sheep population growth. However, the problem with the sensitivity analysis is that the values of sensitivity analysis cannot be compared among different demographic parameters when different scales or units are used. Therefore, instead, the unit-free elasticity is calculated from the elements of the transition matrix (aij) in A, the population growth rate, and the elements of the right and left eigenvectors (wi and vi) as follows (de Kroon et al. 2000): (17) Where ‹w,v› is the scalar product of the two vectors defined as before. The elasticity analysis estimates the effect of a proportional change in the demographic rates on population growth rates. Elasticities are unit-less, represent proportional contribution to λ and are often considered easier to interpret than sensitivities, particularly, when the matrix elements are measured in different units or operate on different scales (e.g. survival rates and fecundity rates) (de Kroon et al. 2000). Since the elasticities are proportions they all sum to one and this property allows us to assess the proportional contribution of each matrix element to λ. For each column of the matrix A, which corresponds to the individual stage, the elasticity values across the rows can be summed to assess how the different stages in sheep production contribute to the sheep growth rate. Similarly, for each row, the sum of elasticity can be made across the columns

Integrated bio-economic simulation model for sheep production: Ex-ante evaluation of investment opportunities in Ethiopia 9 to assess the relative contribution of the different demographic rates to sheep growth rate (for example, fecundity rates and mortality rates). Thus, the analysis of the elasticity helps to identify the stage where the smallest changes in demographic rates will produce the biggest change in the population growth rate and hence this helps to prioritize intervention efforts. Different scenario analyses can also be conducted to take into account the degree of uncertainty and/or variation in the input parameters in the projection matrix. For scenario analysis different values for demographic rates can be used to reflect uncertainty and variation. It is also important to note that the asymptotic multiplication rate of all or part of the population depends on the female demographic rates hence female dominant flock model (Lesnoff 1999). It is indicated that the male demographic rate affects the sex-ratio in the population and the age structure of the male sheep flock only. The sheep flock projection model used here allows to estimate various outputs from the sheep production over time: commercial offtake of live sheep (and its meat equivalents), milk and manure production by different classes of goat, flock sizes and production systems. In order to measure the productivity, feed requirements and output, the special case of a steady state (or zero growth rate) flock, achieved by adjustment of the offtake rates is used. This is particularly important to obtain an unbiased estimate of the impacts of interventions. The steady-state is obtained by re-parameterization of the population projection matrix A such that the dominant eigenvalue of the re-parameterized matrix A is equal to 1. The parameterizing is done by changing the commercial offtake rates numerically. 4.2. Economic model The economic model to analyse the financial and economic profitability of sheep production ‘with’ and ‘without’ policy and investment interventions is based on a capital budgeting approach. This approach requires a stream of costs and benefits to be reduced to a comparable present worth using the process of discounting as outlined in Gittinger (1982). Baseline assessment of financial profitability The financial viability of sheep production ‘without’ intervention is assessed by sheep flock size classes and production systems using the present value (PV) of annual gross margins (GM) generated from sheep production over the time horizon of 20 years. The GM is given as the difference between the total revenue (TRt) of multiple outputs (meat, milk, or manure depending on the sheep production system considered) generated from sheep enterprise and variable costs (VCt) associated with sheep production for a stable sheep flock size and structure. The major variable costs include: costs of veterinary drugs, medicine and mineral supplements. The sheep management affects the reproduction rates and mortality rates which in turn affects the performance of sheep in different production systems by affecting the rate of sheep flock growth and potential offtakes. The flock growth model is linked to the economic model whereby the offtakes from the flock growth model are monetized. Mathematically, the present value of the gross margin (PVGM) for a given flock size class is given as: (18) Where T is the assumed relevant planning time horizon in years; δ is the discount rate (10% assumed); GMt is the annual gross margin accrued in period t. The GM represents the return to a farm’s household labour, land and capital in sheep production and is considered as a farm gate measure of the gross domestic product (GDP) for goat. In general, the higher the PVGM the better. Furthermore, if the availability of data allows, the net profit for sheep production can be obtained as the difference between the GM and the fixed costs. The net profit indicates the extent to which the sheep producers are earning normal or excess profits and thus reflects the level of competition and risk existing in the sheep sub-sector. However, due to lack of data on the fixed costs for individual farm households it was not possible to compute the net farm income or net profits for the sheep production systems.

10 Integrated bio-economic simulation model for sheep production: Ex-ante evaluation of investment opportunities in Ethiopia The life cycle of livestock production is different by livestock species and production systems due to the difference in the biology and objectives of production in different production systems which affect the length or time horizon of the project. In such situations, when project time horizons are different, for comparing the GM for different livestock enterprises or crop enterprises, it is important to analyse the GM in terms of annual equivalent cash flows. Thus, the annualized present value of gross margin (APVGM) is given as: (19) Once the APVGM is derived it can be combined with the geographic information system (GIS) data and sheep population data to generate the sheep GM at herd or household level for aggregation at different levels. The computed GM can be mapped by assigning it to the mapped sheep population in a given production system or for the whole country as a whole. This approach allows to assess the geographic magnitude and distribution of benefits of intervention and to match their level of interventions and expenditures to potential benefits (Thornton and Herrero 2014). Ex-ante impact assessment Financial profitability The financial feasibility of sheep policy and investment interventions is analysed using a multi-year partial budgeting framework (Gittinger 1982). The partial budgeting implies that only the costs and benefits which differ between the ‘with’ and the ‘without’ policy and investment interventions had to be considered. Therefore, the net incremental cash flow (given as a difference between incremental benefits and incremental costs) from the policy and investment interventions over a given planning time horizon (T) is derived and analysed using three principal financial measures: (1) net present value (NPV), (2) benefit cost ratio (BCR), (3) internal rate of return (IRR) (Gittinger 1982). The net incremental cash flow (CFt) due to the policy and investment interventions at time t is mathematically given as: (20) Where and are the monetary benefits of the ‘with’ and ‘without’ interventions, respectively and and are the costs of the ‘with’ and ‘without’ interventions, respectively. Then, the net present value (NPV) of net incremental cash flows from the policy and investment interventions over a T-year planning time horizon is given as: = (21) Where δ is the discount rate assumed to be 10%; I0 is the initial investment cost to be realized over the period t=0; and other variables are defined as before. The NPV provides a basis on which to determine whether the return on a project will be positive or negative and with which to compare different potential projects. As a rule, projects with positive NPVs are accepted. The annualized present value of net incremental benefit (ANPV) from ‘with’ policy and investment intervention for a given flock class in a given production system is given as follows: (22) Where NPV is the net present value of net incremental benefit and the other variables are as defined as before. The income impact of the intervention is also assed in terms of annual net incremental benefit generated per individual sheep and net incremental benefit per capita. In addition, the benefit cost ratios (BCRs) and the internal rate of returns (IRRs) are also calculated for the net incremental benefit from the interventions. The BCR measures the total financial return for each money invested in

Integrated bio-economic simulation model for sheep production: Ex-ante evaluation of investment opportunities in Ethiopia 11 sheep production and provides a measure of the efficiency with which the limited funds are utilized to generate the realized benefits. The BCR criterion measures the effectiveness of investment and is very important in a situation where several projects are competing for the same limited fund. Based on BCR investment criterion, it is advisable to invest if the BCR is greater than one and the priority is given for the project with higher or highest BCR. The BCR is given by the following formula: / (23) Finally, the IRR of the net incremental cash flow for the ‘with’ intervention situation is obtained. The IRR is a discount rate that, when applied to the future streams of project costs and benefits, produces a NPV of zero. It expresses the returns to the investment in the project as an interest rate. It therefore permits the comparison of the returns to investment in the current project with the returns to investment in other possible projects or to simply investing the funds in an interest-earning bank account. Then, in order to make investment decision, the value of IRR is compared with the cost of capital, for example, the value of IRR greater than the current lending rate indicates that the investment is profitable. In other words, the investment must satisfy the condition that IRR is greater than δ to justify initial investment interventions. This indicates that the degree to which the return to sheep investment intervention is comparable to returns to alternative investment options elsewhere and hence competes for the available fund. Mathematically, the IRR is expressed as: =0 (24) Where CFt is the net incremental cash flow at time t and IRR is the internal rate of return. The idea is to solve equation (24) for the IRR. The IRR is solved by trial and error using Microsoft Excel. Future production–consumption balances The impact of policy and investment interventions in sheep production is also assessed in terms of the extent to which it contributes towards closing the growing future production–consumption gaps for sheep meat and milk in Ethiopia. The computation of current and projected sheep and milk production is made using a deterministic 20-years flock growth model as discussed in section 3. On the other hand, the computation of current and projected sheep and milk consumption requirements is made using the information on income elasticity of demand for sheep and milk, growth in real per capita GDP, and human population. Thus, the projected per capita consumption of sheep meat (or milk) is given as: (25) Where Ct is the projected per capita sheep meat (or milk) consumption for a given year t; C0 is a baseline per capita consumption of sheep meat (or milk); η is the income elasticity of demand for sheep meat (or milk); and γ is the trend annual growth rate of real per capita GDP. Then, the projected total consumption of beef (or milk) for future time period t (TCt) is obtained by multiplying the projected per capita consumption with the projected population (POPt) for that given period of time: (26) In this projection, the human population projection is based on the CSA population projection for medium variant population growth scenario. Thus, the annual trend growth rate for real per capita GDP is obtained by taking the difference between the average GDP and population growth rates over the last seven years (2007–13). The income elasticity estimates used in the projections of meat and milk consumptions are based on estimates by the International Food Policy Research Institute derived from household income, consumption and expenditure survey data for 2004/05 (Tafere et al. 2012). The income elasticity estimate for meat and dairy products (including milk) is 0.939 and 0.420, respectively (Tafere et al. 2012).

12 Integrated bio-economic simulation model for sheep production: Ex-ante evaluation of investment opportunities in Ethiopia 5. Results and discussions 5.1 Results of baseline assessments Agro-pastoral and pastoral production systems: Parameterization of the sheep flock growth model Baseline sheep flock size and structure The baseline sheep flock size and structure for the LG production systems used in calibrating the sheep flock projection matrices are given in Table 1. The average sheep flock size for agro-pastoral systems is 16, while the average sheep flock size for small and medium flock size pastoral system is six and 25 heads of sheep, respectively. The structure of sheep flock is analysed at two levels: globally and by sex. Globally (combining all sexes and stages of sheep growth), the sheep flock in both agro-pastoral and pastoral production systems are dominated by female sheep. The female sheep in agro-pastoral flock accounts for 73%, while the male account for about 27% of the sheep flock. The proportion of female sheep in pastoral small size flock is 82%, while in the case of pastoral medium size flock the proportion of female sheep is 75%. In general, in terms of sheep growth stage-wise the adult sheep are dominant, while sex-wise the female sheep are dominant in the sheep flock for agro-pastoral and pastoral production systems. The age-sex structure of sheep flock in LG clearly indicates the focus of agro-pastoralists and pastoralists on reproduction. Table 1. Baseline initial sheep flock size and structure for agro-pastoral and pastoral sheep production systems Stage class Agro-pastoral Pastoral—small flock size Pastoral—medium flock size Flock Structure Flock size Structure Flock size Structure size Global Intra-sex Global Intra-sex Global Intra-sex Young female (Fj) 3.20 0.20 0.27 1.68 0.28 0.34 4.62 0.21 0.28 Sub-adult female (Fs) 1.44 0.09 0.12 0.54 0.09 0.11 2.42 0.11 0.15 Adult female (Fa) 7.04 0.44 0.60 2.7 0.45 0.55 9.46 0.43 0.57 Young male (Mj) 2.88 0.18 0.67 0.54 0.09 0.50 2.20 0.10 0.40 Sub-adult male (Ms) 0.64 0.04 0.15 0.06 0.01 0.06 0.88 0.04 0.16 Adult male (Ma) 0.80 0.05 0.19 0.48 0.08 0.44 2.42 0.11 0.44 Total 16 6 22 Sheep flock demographic rates The detailed initial stage-specific demographic rates used in calibrating the sheep flock projection matrices for agro- pastoral and pastoral production systems are given in Table 2. The sheep demographic rates varied by sex, stage of growth and production systems. The data on demographic rates include: class age range in years, stage duration in years, annual mortality rates, annual survival rate and annual fecundity rates. It is important to note that the observed mortality rate is similar for male sheep and female sheep but relatively higher for younger sheep as compared to older or mature sheep. The mortality rate for the young female sheep varied from 20–21%, while for juvenile male sheep the mortality rate was 22% for all systems assessed. Due to lack of data, similar levels of offtake rates are assumed for both agro-pastoral and pastoral production systems. The initial assumed annual survival rate is also similar

Integrated bio-economic simulation model for sheep production: Ex-ante evaluation of investment opportunities in Ethiopia 13 across different production systems because of the assumed similar levels of annual mortality and annual offtake. The observed annual fecundity rate among the reproductive adult female sheep is considered very low and varies only from 60–66%. In general, the baseline sheep flock in agro-pastoral and pastoral production systems is characterized by high mortality rate, low offtake rate and low fertility rate. Table 2. Initial stage specific demographic rates used in calibrating sheep flock projection matrices for agro-pastoral and pastoral sheep production systems Stage class Class age range Stage duration Annual Annual offtake Annual Annual in year in year (di) mortality rate rate (oi) survivorship fecundity (mi) rate (si) rate (F) Agro-pastoral Fj 0–0.5 0.5 0.20 0.00 0.80 0.00 Fs 0.5–1.5 1 0.10 0.05 0.85 0.00 Fa 1.5–4.5 5 0.05 0.03 0.92 0.60 Mj 0–0.5 0.5 0.22 0.00 0.78 0.00 Ms 0.5–1.5 1 0.15 0.20 0.65 0.00 Ma 1.5–3.5 2 0.10 0.80 0.10 0.00 Pastoral—small flock size Fj 0–0.5 0.5 0.21 0.00 0.79 0.00 Fs 0.5–1.5 1 0.15 0.05 0.80 0.00 Fa 1.5–6.5 5 0.10 0.03 0.87 0.63 Mj 0–0.5 0.5 0.22 0.00 0.78 0.00 Ms 0.5–1.5 1 0.15 0.20 0.65 0.00 Ma 1.5–3.5 2 0.10 0.80 0.10 0.00 Pastoral—medium flock size Fj 0–0.5 0.5 0.21 0.00 0.79 0.00 Fs 0.5–1.5 1 0.15 0.05 0.80 0.00 Fa 1.5–6.5 5 0.10 0.03 0.87 0.66 Mj 0–0.5 0.5 0.22 0.00 0.78 0.00 Ms 0.5–1.5 1 0.15 0.20 0.65 0.00 Ma 1.5–3.5 2 0.10 0.80 0.10 0.00 Stage-structured population projection matrices The annual Lefkovitch stage-structured population projection matrices for agro-pastoral and pastoral sheep flocks are given in Table 3. The population projection matrices are derived using the demographic rates given in Table 2. The matrix is applied to initial sheep flock to provide sheep flock growth dynamics over time. The parameters of population projection matrix determines whether the sheep flock is declining, held constant or growing over time. The parameters also reflect the state of livestock productivity development and affected by the changes in policy and investment interventions. Table 3. Annual Lefkovitch stage-structured population projection matrices for sheep flocks in agro-pastoral and pastoral sheep production systems Stage at year t+1 Stage at year t Fj Fs Fa Mj Ms Ma Agro-pastoral Fj 0.00 0.00 0.55 0.00 0.00 0.00 Fs 0.80 0.34 0.00 0.00 0.00 0.00 Fa 0.00 0.51 0.92 0.00 0.00 0.00 Mj 0.00 0.00 0.55 0.00 0.00 0.00 Ms 0.00 0.00 0.00 0.78 0.26 0.00 Ma 0.00 0.00 0.00 0.00 0.39 0.10 Pastoral—small flock size Fj 0.00 0.00 0.55 0.00 0.00 0.00 Fs 0.79 0.31 0.00 0.00 0.00 0.00 Fa 0.00 0.49 0.87 0.00 0.00 0.00 Mj 0.00 0.00 0.55 0.00 0.00 0.00

14 Integrated bio-economic simulation model for sheep production: Ex-ante evaluation of investment opportunities in Ethiopia Stage at year t+1 Stage at year t Fj Fs Fa Mj Ms Ma Ma 0.00 0.00 0.00 0.00 0.40 0.10 Pastoral—medium flock size Fj 0.00 0.00 0.57 0.00 0.00 0.00 Fs 0.79 0.33 0.00 0.00 0.00 0.00 Fa 0.00 0.47 0.87 0.00 0.00 0.00 Mj 0.00 0.00 0.57 0.00 0.00 0.00 Ms 0.00 0.00 0.00 0.78 0.27 0.00 Ma 0.00 0.00 0.00 0.00 0.38 0.10 Simulation of baseline sheep flock growth dynamics Sheep flock growth rates, stable stage distributions and stage-specific reproductive values The baseline stable stage distribution, stage-specific reproductive values and annual long-term sheep population multiplication rates for agro-pastoral and pastoral and pastoral production systems are given in Table 4. The projected number of sheep by different classes of sheep for agro-pastoral production systems are given in Figures 1, 2, and 3. There is upward trend in the number of sheep for all sheep types and increase is especially noticeable for female adult sheep. Under the baseline scenario, the long-run sheep multiplication rate for agro-pastoral production system is 1.16 which implies a 16% sheep growth rate per year. The annual growth rate for small and medium pastoral production system is 10% and 11%, respectively. The stable stage distribution associated with this constant sheep growth rate is 16%,16%, 33%, 16%, 14% and 5% for juvenile female, sub-adult female, adult female, juvenile male, sub-adult male and adult male sheep, respectively. The stable stage distribution is also dominated by female sheep, especially adult females which accounted for 33% of sheep flock size. This might be because more adult female sheep are required to represent and maintain the flock given high mortality rate. It also observed that the adult female contributes most to the sheep reproduction followed by sub-adult female. The reproductive value of male sheep is assumed to be zero in this kind of sheep growth modelling. Similar stable stage distribution and stage-specific reproductive values are observed for both small and medium size flock pastoral production systems. For example, the adult female is small and medium pastoral sheep flock is 32% and 31%, respectively. Table 4. Baseline stable stage distribution (w), stage-specific reproductive values (v) and annual long-term population multiplication rate for agro-pastoral and pastoral sheep production systems Sex-stage class Agro-pastoral Pastoral - small flock size Pastoral - medium flock size w V w v w V Fj 0.16 1.00 0.16 1.00 0.16 1.00 Fs 0.16 1.45 0.16 1.40 0.16 1.41 Fa 0.33 2.32 0.33 2.29 0.31 2.35 Mj 0.16 0.00 0.16 0.00 0.16 0.00 Ms 0.14 0.00 0.14 0.00 0.15 0.00 Ma 0.05 0.00 0.05 0.00 0.06 0.00 Λ 1.16 1.10 1.11 Figure 1: Number of sheep by sex and stage of growth, agro-pastoral herd Number of animals 100 80 60 40 20 0 2013 2019 2025 2031 Year Fj Fs Fa Mj Ms Ma Source: Simulation results.

You can also read