Determinants of Urban Household Poverty in Arsi Zone, Oromiya, Ethiopia

Transcription

1 ISSN 4-846X An Internatonal Peer-revewed Journal Vol.6, Determnants of Urban Household Poverty n Ars Zone, Oromya, Ethopa Beshr Shaku Berso 1 Adem Kedr Belaneh Legesse 3 1.College of Computng and Informatcs, Haramaya Unversty, Ethopa.Ars Unversty, Ethopa 3.Haramaya Unversty, Ethopa Abstract The man objectve of ths paper s to assess determnants of urban household poverty n urban settng wth a case study from Ars admnstratve Zone, Oromya, Ethopa. Prmary data were collected from 174 households, selected randomly from four towns. Data were analyzed usng descrptve statstcs, one way analyss of varance and multple lnear regresson model. Results show that ncome poverty was postvely nfluenced by educatonal level, household sze and busness partcpaton status of household heads. It was found that ncome poverty was negatvely affected by age of households, martal status and economc status of parents. It was also found that ncome poverty was hgher among dvorced and wdowed household heads as compared to the marred groups. However, ncome poverty was lower for those partcpatng n dfferent busness actvtes than household heads who do not partcpate n busness actvtes. Keywords: Ars, Ethopa, Determnants, Oromya, Urban Household, Poverty. Introducton Poverty s usually understood as the lack of fnancal capacty to purchase the basc human needs but dfferent scholars defne t beyond havng mnmum ncome. It s also defned as a human condton characterzed by sustaned or chronc deprvaton of resources, capactes, choces, securty and power necessary for an adequate standard of lvng and other cvl, cultural, economc, poltcal, as well as socal rghts (Makoka and Kaplan, 005; Adem, 013). Moreover, poverty s not a smple concept. It ncludes hunger, lack of shelter, beng sck and beng unable to see a doctor, not havng access to school and not knowng how to read and wrte, not havng a job, fear for future, losng a chld to llness brought about by unclean water, powerlessness and lack of representaton and freedom (World Bank, 005 ; Sddqu, 009, Adem 013). It can nvolve not only the lack of necesstes of materal wellbeng (the materal requste of lfe) but also the denal of opportuntes for lvng a tolerable lfe (Raj et al., 007). Poverty measures should take nto account socal njustce and other aspects that make one deprved of many rghts (Sahl, 010). Not denyng ths fact poverty s multdmensonal problem, the current study s lmted to ncome poverty only because of budget and tme constrants faced durng the survey tme. The most commonly accepted defnton s related to an ncome based approach whch states that poverty s the lack of ncome or fnancal resources to satsfy the ndvduals basc needs and/or to acheve a mnmum standard of lvng (Snger, 006). Besdes ths, Robnson (011) stated that a strategy n eradcatng poverty among the urban households n busness sector does not only am at ncreasng the ncome level, but also ncreasng overall number of entrepreneurs n a country. Robnson also revealed that urban poor households must be aded not only n terms of busness captal, but also n terms of motvatonal and skll orented tranng that nculcates entrepreneurshp values to be utlzed n commencng soco-economc developments that llustrates the mportance of entrepreneurshp orented human development. Although there may be some poverty studes both n urban and rural context n the country, lttle has been done n Ars Zone to dentfy determnants of urban poverty, especally n relaton to entrepreneural partcpaton. Adem (13), for example nvestgated the rural poverty n Ars zone, but has not consdered the urban cases. Ths study was, therefore, desgned to fll such a gap. The result of the study s beleved to serve the polcy makers to solve the exstng ncome related problems and plans to prevent such problems from happenng n the future. Added to Adem (013), t may also gve a sort of overall poverty n the zone (both rural and urban). It s also mportant for further research n urban poverty and creatng awareness about urban poverty. Wth ths bref ntroducton, the next dscussons n the paper are structured as follows. The methods of data collecton and data analyss are gven n the part two. The thrd part of the paper summarzes the major fndngs followed by dscusson of the fndngs n part four. Fnally, part fve presents conclusons and recommendatons based on the fndngs of the study. MATERIALS AND METHODS Method of Data Collecton Ths study was undertaken n Ars zone of the Oromya Regonal State, Ethopa. Ars zone s found n the central part of Oroma. The zone s dvded nto 5 dstrcts of whch one s the Asella town. The study appled multstage samplng procedure. In the frst stage, four towns namely Asella, Sagure, Dkss and Tcho were randomly selected 106

2 ISSN 4-846X An Internatonal Peer-revewed Journal Vol.6, from 5 towns. Data were then collected from Admnstraton offces of the selected towns on the characterstcs of the urban kebeles under them. On the second stage a total of 9 kebeles were selected randomly. Fnally, 174 households were selected for the ntervew based on smple random samplng. Sample sze of the three towns (Dkss, Sagure and Tcho) was determned usng the formula, n= N Where, C D= Z α / N D +, ( A = P 1 P) A v N A Sample sze for Asella town was determned usng smple random samplng formula, Z n= C α/ pq (1) () (3) (4) Where c s some margn of error to tolerate n estmaton; p s the proporton of poor household; q s the proporton of non poor household; N s total number of households for three towns; n s sample of ( n household for three towns ) 1+ n + n3 and n of Asella town; Z s the value of standard normal dstrbuton for a gven level of sgnfcance (α ); and v s the proporton of populaton of town to the total populaton of household n the selected towns for three towns. In fxng ths sample szes C = 0. 07and C = for three α = 0.05, =0. 0 towns and Asella town, respectvely at P were used. The selecton of p s based on the proporton of households consdered by Adem (013) n hs study on ncome poverty as there s no other related study conducted n the selected area. Ths proporton was supposed to approxmate the proporton of poor n populaton, at least for settng the sample sze for the three towns and Asella town. Accordng to CSA (007) the average household sze of Ars zone s 4.87 persons. By dvdng the total number of predcted populaton of towns to the estmated household sze (4.87) we can obtan the approxmated numbers of household sze. The researcher preferred to use ths average household sze snce there was no detal nformaton about current average household sze n Ars zone. Prmary data were collected (n January-February 015) through personal ntervews of the households and use of structured questonnare wth experenced and traned enumerators. The enumerators who know Englsh language and wth educaton levels of dploma up to frst degree were recruted and traned on how to work n the survey. Data were collected under drect nvolvement and close supervson of the researcher. Secondary data were also collected from the study kebeles and towns admnstraton offces and other related offces. Method of Data Analyss The method of data analyss used for ths partcular study were descrptve statstcs for descrbng general characterstcs of the households, one way ANOVA used to make comparsons between dfferent groups of households wth respect to the characterstcs under consderaton and multple lnear regresson model to assess the determnants of ncome poverty. The dependent varable n the regresson equaton s monthly ncome of households, whch s contnuous. The explanatory varables ncluded n the regresson model can be categorzed as demographc, economc and socal varables. The demographc varables nclude sex of household head, age of household head and martal status of household head. The nsttutonal and ntal condton varables nclude parental economc background. The socal characterstcs nclude relgon. 107

3 ISSN 4-846X An Internatonal Peer-revewed Journal Vol.6, Table 1. Sample sze detal of selected towns No Town Populaton (007) Populaton (015) 1 Number of Households (Approxmate) Sample Sze (households) Sample Sze of Busness- Partcpants Sample Sze of Non- Busness Partcpants 1 Dkss 6,98 8,776 1, Sagure 1,017 15,105 3, Tcho 4,958 6,3 1, Asella 67,69 84,555 17, Total Source: Own Computaton result except for Populaton data generated by CSA (007) Note: The populaton forecast for 015 was obtaned usng.9 percent growth rate (CSA, 007) for Oromya and the geometrc growth model (Adem, 009). Table. Proposed determnants of Income poverty wth the drectons of ther nfluences Monthly Income Expected drecton of nfluence Household sex Not dfferent for both sexes Household head age Negatve Educatonal level of household head Postve Household sze Postve Relgon Not dfferent for dfferent relgons Martal status Monthly ncome lower for dvorced and wdowed Busness partcpaton status Monthly ncome better for partcpated ones. Economc status of parents Better for rch groups parents Multple Lnear Regresson Model Multple regressons are a type of regresson n whch we have a dependent and two or more ndependent varables. The dependent varable s contnuous and the ndependent varables may be quanttatve or qualtatve (category varables). The model for a dependent varable,y, wth observed value q ndependent varables x 1, x,..., x Y = β + β X β X + ε q q q wth observed value (5) x y, y,..., y 1 n (where n s the sample sze) and 1, x,... xq, =1,,...,n Where, Y X s the dependent Varable; are explanatory varables, =1,,..., n β ; 0 β s the constant term; s the coeffcents for a gven explanatory varable ε, and s th random error term (dsturbance term). The term ε s the resdual or random error for ndvdual and represents the devaton of the observed value of the response for ths ndvdual from that expected by the model. These error terms are assumed to have a normal dstrbuton wth mean zero and varanceσ. Thus, ε = Y Yˆ s: s normally dstrbuted wth mean zero and varance σ. The assumptons of multple regresson are: - Independent varables are strctly assumed to be fxed. - Independent varables can nclude contnuous, bnary and categorcal varables. - Addtvety and lnearty: The regresson model s that ts determnstc component s or the expected value of the dependent varable s a lnear functon of the separate predctors. - The error terms are uncorrelated (no seral correlaton). - Equal varance of errors: These random error terms have constant varance. - Normalty of errors: The regresson model assumes that the random error terms are normally dstrbuted wth mean 0 and varance constant. - No multcollnearty: The regresson model assumes that there s no multcollnearty n the data. 108

4 ISSN 4-846X An Internatonal Peer-revewed Journal Vol.6, ε The least squares estmate (OLS) s also the maxmum lkelhood estmate f the errors are ndependent wth equal varance and normally dstrbuted. In any case, the least squares estmator of a vector of lnear regresson coeffcents β s gven by: ˆ 1 β = ( X' X) X' Y (6) In practce, the computaton s performed usng varous effcent matrx decompostons wthout ever fully computng X' X or nvertng t. For ths study, t s merely useful to realze that sβ a lnear functon of the outcomes y consderng the predctors X s a lnear combnaton of the data. The varaton n the dependent varable can be parttoned nto a part due to regresson on the ndependent varables and a resdual term. The latter dvded by ts degrees of freedom (the resdual mean square) gves an estmate of σ and the rato of the regresson β β β mean square to the resdual mean square provdes F-test of the hypothess that 0, 1,..., q takes the value zero. Indvdual regresson coeffcents can be assessed by usng t-statstcs, the rato: ˆ β t = (7) SE( ˆ β) The presence of multcollnearty among the varables serously affects the parameter estmates of any regresson model. The Varance Inflaton Factor (VIF) technque employed to detect the problem of multcollnearty for the contnuous varables (Gujarat, 004). VIF can be defned as; 1 VIF( X j) = (8) 1 R j Rj X s the squared multple correlaton coeffcent between j and other explanatory varables. A larger Where value of VIF ndcates the presence of multcollnearty among varables. As a rule of thumb f a VIF of a varable exceeds 10, the varable s sad to be hghly collnear wth explanatory varables. RESULTS Demographc Characterstcs of the sampled Households Results of the study show that out of the 174 sampled households, 69% are male headed and 31% are female headed. The dstrbuton of the households by martal status shows that 54% of them were marred, 5% were sngle, whle 1% were ether wdowed or dvorced. The average household sze n the study area s.89 wth standard devaton of The mean age of the household head s 34 years wth standard devaton of The ages of the household heads range between 3 and 65 years. The ethnc composton of the sample households ncludes 46% Oromo, 4% Amhara and 1% belongs to other ethnc group. On the other hand, dstrbuton of relgon sample households shows that 56% are Orthodox, 3% are Muslm, 0% are Chrstan Protestant and 1% belongs to other relgon groups. Determnants of Income Poverty Income poverty ndcators consdered n ths study nclude sex of household head, age, educatonal level of household head, household sze, relgon, martal status, busness partcpaton status, economc status of parents and parental economc background. Fnally, the model output of the determnants of household ncome poverty are gven. The survey results show that the monthly ncome of household heads was sgnfcantly dfferent among martal status groups (marred, sngle, dvorced and wdowed) or the martal status groups are not nfluenced by ncome poverty equal. Smlarly, the ncome of household heads was sgnfcantly dfferent between economc status of parents groups (rch, medum and poor) and results also reveals that monthly ncome of household head was not sgnfcantly dfferent among relgon of households and ethncty groups (Table 4). The results of the regresson analyss, on the other hand, show that among the proposed explanatory varables for affectng monthly ncome, only age, educatonal level, household sze, martal status (dvorced and wdowed), economc status of parents (medum, poor) and busness partcpaton status (partcpated) were found to be statstcally sgnfcant n ths study (Table 5). Accordngly, age of household head negatvely determne the ncome of the household head whch mples that younger households generate more ncome than old aged household heads. In addton, educatonal level of household head has a postvely sgnfcant effect on the monthly ncome of households at 10% level of sgnfcance. Smlarly, household sze and partcpatng n entrepreneural 109

5 ISSN 4-846X An Internatonal Peer-revewed Journal Vol.6, actvtes were postvely related wth monthly ncome of households. Table 3. Households Demographc Characterstcs Demographc characterstc Category/measure Value Household sex Male 69% Female 31% Age of household head n years Mean Standard devaton Household sze Mean Standard devaton Ethncty Oromo 46% Amhara 4% Others 1% Relgon Muslm 3% Orthodox 56% Protestant 0% Marred 54% Martal status Sngle 5% Dvorced and Wdowed 1% Source: Survey Data Table 4: One way ANOVA test among monthly ncome of households and dfferent groups Varables Monthly ncome of household heads Sum Squares df Mean Squares F Sgn Ethncty.15E E E E +6 Relgon 7.49E E E E +6 Martal Status 6.95E E E E Economc status of 3.68E E parents 7.39E E+6 Source: Survey Data Table 5: Determnants of household monthly ncome. Varables Coef. St.err T P>/T/ Sex of household head: 1= male Age of household head Educatonal level of household head Household sze Martal status: = Marred 3=Dvorced 4=Wdowed Partcpaton status: 1=Partcpated Relgon of household head: =Orthodox 3=Protestant Economc status of parents: =Medum 3=Poor Constants Note: Number of obs =174 F (1, 161) = 7.38*** Prob > F= R-squared= , Adj R-squared = , Level of sgnfcance s at α = 5% and 10% The ncome of dvorced and wdowed households was less as compared to sngle household heads. It was also found that the ncome of household heads who come from medum and poor parents s less as compared to households who come from rch parents and the ncome of household heads who partcpated n dfferent busness actvtes more as compared to households who do not partcpated. DISCUSSIONS These fndngs are n agreement wth what have been reported by Gan (007), Masood and Nasr (010) and 110

6 ISSN 4-846X An Internatonal Peer-revewed Journal Vol.6, Ataguba et al (01): as they all argued that famly sze and educatonal level are most mportant determnants to reduce poverty. But the current study contradcts wth Gan, Masood and Nasr, and Ataguba wth regards to other sgnfcant determnant varables: Such as, sex of household head, age of household heads, martal status, busness partcpaton status and economc status of parents. Smlarly, ths study result s n agreement wth Adem (013) whch ponted out that educatonal level of household heads, age of household heads, and economc status of parents were the most mportant varables for ncome poverty. But, contradcts wth the varables such as household sze, martal status, and busness partcpaton status. In lne wth ths, fndngs of the present study are n agreement wth Adofu (013) whch ndcated that educatonal level of household heads play mportant role to reduce poverty. But the current study ndcates that not only educaton of households, but also age of household heads, household sze, busness partcpaton status, economc status of parents and others are mportant determnants for poverty reducton. Abdel and Yasn (013) revealed that partcpatng n small enterprses could play mportant role n generatng more ncome and Paul (006) argued that small enterprses make a contrbuton to reduce poverty by creatng employment and generate ncome for themselves and those they hre. Fndngs of the current study are n agreement wth the dea of Abdel and Yasn (003) and Paul (006). But, somewhat contradct the dea of Paul, snce ths study clams that not only small enterprses but also partcpatng n dfferent entrepreneural actvtes plays an mportant role to generate more ncome. CONCLUSIONS AND RECOMMENDATIONS Based on the fndngs of ths study, the followng conclusons are made. Frstly, younger household heads do generate more ncome than old household heads whch mples that as age of households ncrease the ablty to generate more ncome decreases. Secondly, Educatonal level of household heads s the most mportant component for ncome of households. Household sze s also an mportant varable to reduce ncome poverty whch mples that large famly generates more ncome by partcpatng n dfferent works and reduces poverty. Thrdly, partcpatng n dfferent entrepreneural actvtes s a very essental to generate more ncome. Fourth, dvorced and wdowed households are found to generate less ncome as compared to sngle groups. Generatng less ncome may not allow the dvorced and wdowed households to concentrate on dfferent busness actvtes or other works as requred and they are more affected by ncome poverty. Households who come from medum and poor parents are more affected by ncome poverty as compared to those comng from rch group parents. Based on the fndngs of ths study, the followng recommendatons or polcy mplcatons are made: Frstly, there s need to gve attenton on the households educaton. Therefore, the relevant authortes should develop programme that gve awareness on educaton of households. Secondly, there s need to gve specal attenton to dvorced and wdowed households. So, t s better f government or concerned body gve specal attenton to dvorced and wdowed households. Ths may nclude awareness creaton, fnancal support or creatng good condton/atmosphere for them. Thrdly, there s need to ncrease the number of partcpants n busness actvtes. So, the government or relevant authortes should ncrease the capacty of entrepreneurs through awareness creaton, fnancal support and other facltes. Fnally, further studes recommended that researchers or polcy makers have to see the contrbuton of partcpaton n entrepreneural actvtes on mult-dmensonal poverty, nstead of ncome poverty n the future. REFERENCE Adem Kedr Geleto Introducton to Statstcs and Its Applcaton, nd edton. Adem Kedr Geleto Determnants, Consequences and Copng strateges of Rural Poverty n Ars zone, Oromya, Ethopa, PhD Dssertaton, Haramaya Unversty, Haramaya, Ethopa. Apata,T.G Determnants of rural poverty n Ngera: Evdence from small holder farmers n South-western Ngera, Journal of Scence and Technology Educaton Research, 1(4): Ataguba, K. 01. Determnants of Multdmensonal Poverty n Nsukka, Ngera. Barja, G. and B.S. Ggler The Concept of Informaton Poverty and How to Measure t n the Latn Amercan Context, Dgtal Poverty: Latn Amerca and Carbbean Perspectves: 11-8, Workng paper MPD 006/005. C.Gan.007. Determnants of Urban Household Poverty n Malaysa. Journal of Socal Scences, 3(4): Central Statstcal Agency (CSA) Ethopan Populaton and Housng Census, provsonal Result for 007 by Urban and Rural, CSA, Adds Ababa. Cochran, G. W. 00. Samplng Technques, 3 rd edton, Wley, Delh. G.S. Maddala Introducton to Econometrcs nd edton. Hopkns, J., D. Southgate and C. Gonzalez-Vega Rural poverty and Land Degradaton n El Salvador, n Amercan Agrcultural Economcs Assocaton Annual Meetng. J. Akael Determnants of Rural ncome n Tanzana: An Emprcal Approach Research Report 10 (4). Koutsoyanns, A Theory of Econometrcs: An Introductory Exposton of Econometrc Methods, Palgrave, 111

7 ISSN 4-846X An Internatonal Peer-revewed Journal Vol.6, Ontaro. Makoka, D and M. Kaplan Poverty and Vulnerablty: Term Paper for Interdscplnary Course Internatonal Doctoral Studes Programme. Masood, S. A. and Nar, I Determnants of Urban poverty: The case of medum sze cty n Pakstan. Msturell, F and Heffersan, C What s Poverty? A Dachromc Exploraton of Dscourse on poverty from the 1970s to 000s: European Journal of Development Research, 0(4): Nasr Iqbal Determnants of Urban Poverty: The Case of Medum Sze Cty n Pakstan. Noorderhaven, N The Role of Dssatsfacton and Per Capta Income n explotng self-employment across 15 countres, Entrepreneurshp Theory and Practce, 8(5): O Connor, A The Study of Poverty n Afrca, Belhaven Press London. Owen Barder What s Poverty Reducton? CGD Workng paper 170. Washngton, D.C. Center for Global Development. Raj, F Contemporary Development Economcs, From Adam Smth to Amartya Sen, New Central Book Agency, Kolkata. Sahl, I.M.G Challengng the Concept of Poverty: Does Islam provdes a Soluton? Socetal Studes, Sddqu, M Determnants of Poverty n Pakstan: Fndngs from Survey Data 005, European Journal of Socal Scences, 1(1): Taylor Statstcal Methods, nd edton. World Bank Introducton to Poverty Analyss, [Onlne], World Bank Insttute, Avalable from: Accessed on January 05,