Research Thesis defence ppt_Anteneh!.pptx

Research Thesis on Poverty in Urban area using dummy variable of STATA software.

1. Determinants of Poverty in Urban Households (The Case of Aysiata District in Afar Region, Ethiopia)   By     Anteneh Tadele

2. Determinants of Poverty in Urban Households A research thesis submitted to the college of Business and Economics department in partial fulfilment of the requirements for the Degree of Master of Science in Developmental Economics   Major Advisor: Mohammed Adem (Asist. Prof) Co – advisor: Abdulrezak Nejmuhdin (MSc) February 2024 Samara University, Ethiopia

3. Acknowledgement First, I praise the Almighty God who help me to reach this end. Second , I would like to record a special note of thanks to my research Advisor Mr. Mohammed Adem (Assist. Prof.) and co – advisor Mr. Abdulrazak Nejmudin for (MSc) for his earnest and constructive comments throughout the analysis and to the final write-up of the thesis by adding valuable comments on realization of the thesis. Third, I would like to thank my previous employer, SCI for sponsor me to learn this MSc program. I am very thankful to Aysaita Woreda Agriculture and Rural D evelopment and Health O ffice for supporting given real information to find the total number of households and livestock in the district. Finally, my special thanks go to my beloved wife Mushira Wondie , she is always supporting and helps me overcome difficulties without complaining .

4. Outline Background of the Study Statement of the Problem General and Specific Objective, Research Questions Research Methodology [ Data Type, Sources, Analysis] Result and Discussion Summary and Conclusion Recommendation

5. 1. Background of the study The study aimed to describe the determinants of poverty urban household in Aysaita district in Afar region, Ethiopia. The focus was how households can improve income to out of poverty line through successful knowledge-sharing using CBN approach. Poverty in the simplest sense of the word is a state where one lack of accessing to basic needs such as food, clothing, and shelter. Poverty is defined access to services and security critical to well-being and not just income and consumption (WV, 2022). According to the World Bank 2000s, 2015, and 2021, the poverty line per day is $1, $1.90, and $2.15/day, respactivily but currently, 689 million people were living in extreme monetary poverty in 2017, WB, 2020 Ethiopia is one of the poorest countries in the world with a large portion of its population believed to be living below poverty line. As shown from WB, 2021 and WFP, 2023 result, t he headcount index of absolute poverty of Ethiopia was decreased from 44.2 percent in 2000 to 23.5 percent, in 2015 and 27.8 percent in 2016 , respectively.

6. Cont.… Afar region is one of the poorest and least developed regions of Ethiopia, neglected by national development efforts. According to regional report by ( BoFED 2020), food security situation in Aysaita district was below normal. MoFED , 2019 report shows that, the h eadcount poverty incidence in Afar region declined from 41.9 percent in 2004/05 to 32.8 percent in 2020/21. In the same period, the rural poverty head count index declined from 45.2 percent to 35.1 percent while urban declined from 35.3 percent to 23.1 percent. According to DRMFSAB, 2020 report, poverty situation in Aysaita district was high, the zone was suffering food related problems. Therefore, to assist the poor in the district it requires identifying the factors determining poverty in locally specified context and need to measure the intensity of poverty in the district.

7. 2. Statement of the Problem Conducting a research on determinant of poverty is crucial for identifying the level of income and expenditure to establish a sustainable competitive usage and improve overall profitability of the household incomes. The conclusion that poor households is positivity correlated with their explanatory variables has both proponents and opponents in the empirical literatures. Bashir (2010) and Sharma (2014) shows that the age of household head is negatively correlated with the probability of being poor. Whereas Indris (2012) found out that age of the household head affects food insecurity positively. Abubeker M/d et al 2014, show that in the rural site of the study area, 52.8% have been living below the poverty line with poverty gap and poverty severity indices of 0.16 and 0.07, respectively and Gini coefficient of 0.31.

8. Cont.… Most research studied on poverty is target individual rather than household level. On the other hand, study not to conduct that much to show the impact of determinants of unidimensional poverty using CBN approach in the district. There is an insignificant number of research works focusing on poverty using unidimensional methods in the Aysaita district – especially in urban area. Adding to this, some variables that can determine poverty are included like Children under 14 years and ownership of cell phone service were used to differentiate others researcher like ( Abubeker M/d, Ayalneh Bogale, and Asefa Seyoum 2013/14 ) conducted in Aysaita in rural area. More Q uantitative methods of data collection were employed, and under U nidimensional than Multidimensional poverty type – CBN approach was used for setting poverty line.

9. 3. Objectives and Research Questions General Objective To identify the determinants of poverty status of urban households in the case of Aysaita District of the Afar Regional State. Specific Objective Identify the poverty status – incident of households. Estimate the level and intensity of poverty of households. Analyze factors affecting poverty status of households . Research Questions What is the poverty status of the respondents? What are the levels of poverty among households? What are the determinants of poverty among households?

10. 4. Methodology of the research The researcher used Primary and some Secondary source of data and more used for Q uantitative data than qualitative using structural questionnaire and its sampling procedure was multi-stage sampling. Finally, using Yemane’s (1971) sample size formula, 288 sample households were selected randomly based on Probability Proportional to Size (PPS) of the population in each kebele. Data analysis used for cross-sectional data from 288 sample and, under Unidimensional poverty type – CBN approach were employed to identify poverty line using Foster-Greer- Thorbecke (FGT) index for descriptive and, Under Limited Dependent V ariable Model [B ecause of the DV is binary 0/1] logit model were used for econometrics analysis to identify determinant of poverty.

11. Model Specifications Limited Dependent Variables model of magnitude/intensity, and determinant of poverty. Structure of the models: E ( y i /x ) = Pr [ y i = 1/x ] = G [β 1 + β 2 X 2i + β 3 X 3i …….+ β k X ki ] Identify Poor households – CBN approach, FGT equation y 1𝑖 = 𝑋 1𝑖 𝛽 1 +𝜀 1……. ..(1) y 1𝑖 = 1 if y 1𝑖 is poor 0 otherwise 2. Determinant of poverty - logit & Marginal effect y 2𝑖 = 𝑋 2𝑖 𝛽 2 +𝜀 2𝑖 ……. (2) Where; y 2i is observed if and only if y 1 𝑖  1 [poor]; the variance 𝜀 2𝑖 is normalized to one because only y 2𝑖 not y 1i . DV : Poverty incidence – which a dummy variable [1- Poor; 0 – non-poor] IV : Sex, Age, Marital status, Edu_level , FamSz_AE , Children under 14, Emp_level , TLU_AE, Own Cellphone, Credit services, Marketplace access and Market Distance .

12. 5. Result and Discussion As we know, unidimensional poverty measurement involves two steps (Sen 1976): Identification - to identify who is poor and, Aggregation - effects how poor the respondents is?. Descriptive Analysis Analysis of D iscrete Variable Variable Mean Std. Dev Min Max Age 39.12 10.91 22 69 FamSZ_AE 3.9 1.8 2 9 Child<14 1.8 1.1 4 Child>51 0.24 0.52 2 TLU_AE 8.43 3.90 18 Percapinc 1545.97 830.82 277.78 4756.50 Mark_Dist 3.54 1.38 1 7

13. Cont.… Analysis of Categorical Variable Variable Category Frequency Percent   Name of Kebele Amolederewa(04) 76 26.39 Kulsi’coma (02) 67 23.26 Beri’daba (01) 86 29.86 Aberoberi’fagi (03) 59 20.49 Sex of the household head Female 93 32.29 Male 195 67.71 Marital status of household head Unmarried 64 22.22 Married 232 77.78 Educational status of household head Illiterate 89 30.90 Primary Education 112 38.89 Secondary Education 65 22.57 Higher Education 22 7.64

14. Conti… Variable Category Frequency Percent Employment Status of household head Unemployed 198 68.75 Employed 90 31.25 Households Having Agricultural area No 87 30.21 Yes 201 69.79 Households Having Livestock Owen No 43 14.93 Yes 245 85.07 Household accessing Market service No 133 46.18 Yes 155 53.82 Households accessing Credit/Loan Service No 227 78.82 Yes 61 21.18 Household having own Telcom or Cellphone No 167 57.99 Yes 121 42.01

15. Cont.… Sources of Income of the Household heads Based on the nature of their livelihoods, households in the district had depend on different sources of income. We found that households participate in a range of types of employment or activities to generate income and maintain themselves. From the respondent household survey result shows that, 38.69 percent of the households do not have extra jobs to improve their monthly or yearly income. Sources of Income Frequency Percent Government employed [recruited] 69 23.96 On-farm (Agricultural product) 49 17.01 Off – farm (Livestock products) 53 18.40 Self-employed: petty trade, charcoal sold 81 28.13 Private Institution employed: 36 12.50

16. Dimensions of Poverty among Urban Households Status of Poverty - Who are the Poor? Using CBN approach , the following steps were employed to obtain the food poverty line. Identify and select 11 food items commonly consumed by most of the poor. Each food item in the bundle of goods is weighed in kilograms and liters. Each unit of the food items is divided by the adult equivalent units-AEU and sum all. 2,200 kcal being the minimum calorie required adult equivalent per day in Ethiopia MoFED , 2020. Finally, a given food expense value was a poverty line threshold that provides a monetary value for the food and non-food component, so t he food expense value was 2,823ETB and from this 23 percent [649.29 ETB] are use the non-food poverty which the share of the lowest expenditure distribution so food poverty line was 3,472 ETB household per month but per adult equivalent is [3,472/4] = 868 birr.

17. Cont.. s/n Food items kcal/100g Kg/month /AEU Kcal/month /AEU Kcal/AEU /Day Share (%) Average price Food expenses 1 Teff 358 10.10 25,310.60 844 0.19 80.0 808 2 Wheat 326 4.3 9,812.60 327 0.11 65.0 280 3 Maize 361 5.2 13,140.40 438 0.20 55.0 286 4 Sorghum 344 1.75 4,214.00 140 0.06 60.0 105 5 Onion 37 2.15 556.85 19 0.01 80.0 172 6 Red paper 311 0.7 1,523.90 51 0.02 180.0 126 7 Meat 289 1.5 3,036.08 101 0.05 450.0 675 8 Sugar &Salt 400 0.5 1,400.00 47 0.02 80.0 40 9 Milk 65 1.21 550.55 18 0.01 80.0 97 10 Oil 900 0.5 3,150.00 105 0.05 350.0 175 11 Coffee &tea 1180 0.4 3,304.00 110 0.05 150.0 60   Total 4,571 28.31 65,998.98 2,200 77% 2,823

18. Dimensions of Poverty among Urban Households FGT index estimates, FGT(a) and Gini Coeff icient Explanatory Variables    Category Head Count (P ) Gap index (P 1 ) Severity (P 2 ) Gini Coefficient Overall index 0.382 0.104 0.032 0.232 Name of Kebele Amolederewa (04) 0.355 0.101 0.032 0.245 Kulsi’coma (02) 0.418 0.117 0.037 0.216 Beri’daba (01) 0.349 0.098 0.031 0.229 Aberoberi’fagi (03) 0.424 0.102 0.028 0.228 Gender of Household Head Female 0.484 0.137 0.044 0.248 Male 0.333 0.088 0.027 0.222 Marital status of Household Head Unmarried 0.672 0.189 0.061 0.229 Married 0.299 0.079 0.024 0.224

19. Cont.. Explanatory Variables  Category P P 1 P 2 Gini Overall index 0.382 0.104 0.032 0.232 Educational Status of Household Head Illiterate 0.736 0.739 0.740 0.156 Primary Education 0.154 0.143 0.133 0.208 Secondary Edu 0.100 0.112 0.122 0.222 Higher Education 0.136 0.045 0.016 0.145 Employment Status or level Unemployed 0.471 0.130 0.041 0.227 Employed 0.159 0.038 0.010 0.226 Accessing Market center No 0.551 0.153 0.048 0.229 Yes 0.227 0.059 0.017 0.216 Accessing Credit or loan Service No 0.440 0.119 0.037 0.235 Yes 0.129 0.036 0.012 0.199 Cellphone Ownership No 0.471 0.134 0.042 0.234 Yes 0.222 0.060 0.019 0.217

20. Status of Income Inequality in District - Lorenz Curve Gini index value, 01 kebele = 0.229 , 02 kebele = 0.216 , 03 kebele = 0.228 , and 04 kebele = 0.245 In the study area, the overall income inequality was 0.232. Results show that the Gini coefficient of 04 kebele is 0.245 while 01 is 0.229 and 03 kebeles is 0.228, therefore, 04 kebele was slightly higher . The relatively high Gini coefficient is in 02 kebele which is 0.216 indicates that unequal distribution of consumption expenditure. Therefore, the closer the line of equality the lower the income inequality.

21. Econometric Analysis Binary logistic regression analysis was used to identify the effect of each independent variable on the studied district poverty status of the households. Before, regress the logistic model, check the goodness of fit in the predicted variables using the Hosmer- Lemeshow test model. Therefore, Logistic model for poor, goodness-of-fit test number of observations = 288 number of groups = 10 Hosmer- Lemeshow chi2(8) = 9.40 Prob > chi2 = 0.3094 The results indicates there is a good logistic regression model because of there is a small chi square with large p-value greater than 0.05. Therefore, there is NO difference between the observed and the model predicted values.

22. Econometric Analysis result Since the logistic model is nonlinear , the marginal effects of each independent variable on the dependent variable are not constant but they are dependent on the values of the independent variables (Green, 1993, cited in J.G. Rodriguez, 2010). The results of the logistic model were given below consisting of the variables, the estimated odds ratio, and the marginal effects for explanatory variables included in the model. The odds are the ratio of the probability of being poor to the probability of not being poor. The odds ratio gives the change in the odds of being poor as opposed to not being poor. It describes the correlation between the dependent and independent variables, i.e., P ositively or Negatively correlated of the probability of being poor to non-poor. The M arginal effect is the percentage change in probability associated with a unit change in the explanatory variable .

23. Logistic & Marginal effect output Logistic regression Number of obs. = 288 LR chi2(15) = 274.27 Prob > chi2 = 0.0000 Log likelihood = -54.385646 Pseudo R2 = 0.7160 Independent Variables Odds Ratio Std. Err. Z P>|z| Marg. Effect Sex of hh head (1. Male) .878 .482 -.24 0.813 -.0074 Age of hh head 1.709 .346 2.64 0.006** .0304 Age_sqr .995 .002 -2.41 0.013** -.0003 Martl_stat (1.Married) .925 .0703 -3.13 0.002** -.1546 FamSZ_AE 1.545 .859 2.12 0.029* .0247 Childunder14 .739 .236 -0.95 0.341 -.0171

24. Cont.… Independent Variables Odds Ratio Std. Err. Z P>|z| Marg. Effect Education .286 .151 -2.36 0.013* -.0710 Illiterate 1.128 .859 0.16 0.875 .0068 Emply_sta (1. Employed) .728 .493 -0.47 0.642 -.1832 TLU_AE .715 .061 -3.95 0.000*** -.0190 Percapinc .999 .0001 -5.29 0.000*** -.0015 Mkt_access (1. Yes) .275 .161 -2.21 0.027** -.0771 Credit_serv (1. Yes) .412 .368 -0.99 0.326 -.0511 Own_phone (1.Yes) .797 .449 -0.40 0.689 -.0129 MktDist .838 .159 -0.93 0.349 -.0100 _cons .745 3.097 -0.07    

25. Cont.… The above table, estimated binary logit regression model and the interpretations for odds ratio are as follows: Age , Illiterate , and FamSZ_AE have odd ratios greater than one, which means that; these variables are positively correlated with the poverty likelihood of being poor households and Sex of hh head , Agesqr , Martl_stat , Childunder14 , Education , Emply_sta , TLU_AE , Percapinc , Mkt_access , Credit_serv , Own_phone and MktDist are negatively correlated with the poverty likelihood of being poor because their odd ratios is less than one. The likelihood ratio Chi-square of 274.27 with a p-value of 0.0000 tells us that our model fits significantly better than an empty model (i.e., a model with no predictors). Out of the total thirteen explanatory variables hypothesized to determine intensity of poverty, seven of them are significant at 1 and 5 percent of statistical level and the rest are not significant. Statistically significant explanatory variables are Age , Agesqr , FamSZ_A , Martl_stat , Education , TLU_AE , Percapinc , and Mkt_access .

26. Cont.… The result shows that the model's outcome shows that the Sex of Household Head (OR: 0.878) this means that a household led by a man has a 0. 878 times lower chance of being unidimensional poor than a home headed by a female headed household. In other word, female-headed households had a lower probability of being poor as compared to male-headed households. The negative related but not significant association between the variable sex and intensity of poverty among the district shows that male-headed households in the area were vulnerable to poverty than female-headed households. This shows that having a male-headed household poverty intensity of poor households decrease by .074 percent. The magnitude of the family size of household was positively, and statistically affect at 1 percent significance level. On average, one unit increase in family size (in AE) among poor households increases the intensity of poverty by 1.55 percent if all other variables are held at their mean value.