ASSESSING THE WELFARE INCOME INEQUALITY IN EDO STATE

Transcription

1 ASSESSING THE WELFARE INCOME INEQUALITY IN EDO STATE SUMMARY OSUNDE OMORUYI; OSASERE GREG IGBINOMWANHIA Department of Socology and Anthropology Unversty of Benn, Benn Cty E-mal: And EFOSA ODIGHIZUWA; HELEN EHI EWEKA Department of Socal Work Unversty of Benn, Benn Cty We examned n ths paper the changes n household earnngs due to demographc change. We used a number of nequalty measures to dentfy the level of ncome nequalty at dfferent ponts of the dstrbuton of ncome usng the Natonal Lvng Standard Survey From our estmaton, we observed that educaton plays a sgnfcant role n hgher average household earnngs among households wth wage employment. Analysng the contrbuton of ndvdual determnstc factor to ncome nequalty, we further observed that farm ncome s the greatest factor towards overall nequalty for most household group n Edo State. INTRODUCTION Ths study examnes the dstrbuton of personal welfare wthn the populaton of Edo State. The queston frequently asked by polcy makers over the past 25 years s how s nequalty generated (Clarke; Coln and Zou 2003? In ths study we ntend to fnd out ssues responsble for ncome nequalty such as returns to educaton, ndvdual personal characterstcs (endowments, ther labour market partcpaton and the composton and sze of the famly. Ths analyss uses a number of measurements of nequalty so as to measure the level of nequalty at dfferent ponts of the dstrbuton of ncome n Edo State. Snce the ntroducton of Structural Adjustment Program n 1986 n Ngera, the level of nequalty has ncreased consderably. As Sala--Martn and Subramanan (2003 ponted out Ngera has experenced a sharp reducton n ncome dstrbuton: n 1970 the top 2 percent and the bottom 17 percent earned the same amount of ncome n Ngera, but by 2000, the top 2 percent had the same ncome as the bottom 55 percent. At ths pont the dspartes n GDP per capta show sgnfcant nequalty between the sx geo-poltcal zones n Ngera.

2 The GDP per capta n the North stood at $718 and that of the South-West was $1,436, whle that of the South-South and South-East stood at $2,010 and $933 respectvely (Rewane Therefore, nequalty between dstnct areas wthn Ngera s a major ssue for polcy makers. On the other hand, surprsngly lttle attenton has been gven to the dstrbuton of ncome wthn dfferent states n Ngera, especally n Edo State. It s the am of ths study to fll ths gap n the development economcs lterature. However, studes already conducted on nequalty n Ngera are mostly based on consumpton as a welfare ndcator, whle we wll focus on ncome as a welfare ndcator for assessng the level of nequalty n Edo State. Varous studes have been carred out on the dfferences n wage dstrbutons, through methods such as nequalty decomposton poneered by Oaxaca (1973 and Blnder (1973. Ther studes attempted to determne the contrbutons of socoeconomc varables to ncome nequalty. Bourgugnon; Ferrera and Lete (2002 extended ther framework by usng a decomposton method based on a parametrc representaton to examne the changes n soco-demographc structure of the populaton on nequalty n household ncome dstrbuton n Brazl, Mexco and US and found out that Brazl s nequalty was due to poor access to assets and transfers that potentally generate non-labour ncome e.g. penson relatvely compared to Mexco and US. 1 As asserted before, the level of nequalty has been on the ncrease especally n developng countres lke Ngera. Addson and Corna (2001 and Kanbur and Lustg (1999 ponted out that ncreasng ncome nequalty has a negatve mpact on ndvdual welfare. Ther asserton was n consensus wth Brdsall; Ross and Sabot (1995, 1997 that ncreasng nequalty has a negatve mpact on growth operatng va channels such as asset endowments, savngs rate, nvestment n educaton and expectaton. Measurng welfare nequalty has both conceptual and emprcal challenge. Access to mcro data n analysng the evoluton of welfare nequalty overtme n Ngera especally at the regonal level s another major problem. Detaled studes have examned the mpact of nequalty on aggregates growth outcomes and explored the possble ways, such as ncentve for workng and savng. These studes suggest that nequalty can have perverse effects on aggregate welfare. Ther results are nconclusve; for nstance Barro (2001 argues that nequalty s bad for economc growth for countres wth per capta GNP s below $500 but good for growth for countres wth GNP s above that level. Other studes conducted on household nequalty n Ngera compare rural and urban household nequalty usng ncome ndcator. For example, Oyekale, Adeot and Oyekale (2006 observed that ncome nequalty s hgher n rural areas than urban areas n Ngera. Adams and Jane (1995 also examned the sources of

3 ncome nequalty and poverty n rural Pakstan usng ncome data. They dscovered that two ncome sources nonfarm and lve-stock tend to reduce overall nequalty, whle the other three sources such as agrculture, transfers and rental wll ncrease overall nequalty. A paper by Bhalla (2003 reported that both urban and rural Gn coeffcents declned between and Accordng to hs calculaton, rural nequalty decreased n 15 out of 16 major states n Inda, and urban nequalty reduced n 8 of the 17 states over ths perod. Contrarly, regonal dspartes are hgh and ncreasng especally n Afrcan contnent. For example Sahn and Stfel (2003 n Afrca; Escobal and Torero (2003 n Latn Amerca; and Fredman (2005 n Asa; reported n ther varous studes that regonal nequaltes are hgh, and they have been rsng. Frawley; Commns; Scott and Trace. (2000 assessed the factors underlyng the emergence of the low-ncome farm households and the effectveness of farm ncome and general welfare polces n counterng the lack of vablty of the farm sector. They found that low-ncome farm households were dstrbuted evenly throughout the state and were found to be closely assocated wth small farms and dry stock farm systems. They also found however that those low ncome farm households are no more deprved than the average of all households n the state. Awoyem and Adeola (2004 used the standard Gn decomposton approach to look at the sources of nequalty n rural Ngera. Ther result shows that agrcultural ncome contrbuted to most nequalty to total ncome. As one can see from the studes cted n ths paper, regonal ncome nequalty studes especally at the state or provnce level have receved far less attenton n recent years. The am of ths study s to fll the gap. In ths study we wll use the ncome data as ndcator to look at welfare nequalty n Edo state. We apply n ths study some contemporary and current analytcal approach to decomposed ncome nequalty n Edo state. BRIEF DESCRIPTION OF EDO STATE Edo State s located n the South Southern part of Ngera. It was carved out of the then Bendel State n Whle the State can be sad to be homogenous wth all the varous ethnc groups such as the Benn, the Esan and Afema, the Akoko-Edo and the Owans sharng dentcal hstorcal orgn and poltcal nsttuton, dfferences exst especally n the area of culture among the Etsako and Akoko-Edo people. The word Edo however refers to those peoples who have shared hstorcal orgn as well as poltcal and cultural smlartes. These peoples are called the Edod people. The state shares boundares wth Delta on the south, Ondo on the west, Kog on the North-East. The man town n the state are Benn, captal of the ancent Benn Kngdom whch s also the state captal, Ubaja, Auch, Ekpoma and Urom. Furthermore, the most predomnant occupaton of the people n Edo state s agrculture. The major cash crops produced are rubber, cocoa and palm produce. In addton, the state produces such crops as yam, cassava, rce, plantans,

4 gunea-corn and assorted types of fruts and vegetables. Edo state s also endowed wth abundant natural resources. The prncpal mneral resources nclude crude ol, natural gas, clay chalk, marbles and lmestone. In the aspect of educaton, the state has such educatonal nsttutons as the Unversty of Benn, Edo State Unversty, and Auch Polytechnc among others. The state s known for ts educatonal attanment as t s the home of many dstngushed scholars. It also has such tourst attractons as Oba s Palace n Benn Cty, Ramat Park also n Benn Cty and Sakpoba Holday Resort. A total of 2.86 mllon persons are lvng n Edo State. Ths comprses of 51 percent males and 49 percent females (USAID, 2002; and NPC The State s made up of eghteen (18 local government areas from varous ethnc groupng as mentoned earler as shown n Table 1 below: Table 1: Local Government Areas n Edo State 1. Akoko-Edo 1. Esan Central 2. Esan North East 3. Esan South East 4. Esan West 5. Etsako Central 6. Etsako East 7. Etsako West 8. Igueben 9. Ikpoba Okha 10. Owan West 11. Owan East 12. Oredo 13. Orhonmwon 14. Ova North East 15. Ova South 2. Egor 16. Uhuomwode West Source: METHODOLOGY In ths paper, our objectve s to dentfy the sources of nequalty and ts demographc patterns n Edo State. By demographc patterns, we mean the number of household, wth partcular demographc characterstcs,.e. number of chldren per household, number of elderly per household and the martal status of adults wthn the household and how these affect household earnngs. Furthermore, we then descrbe the system of equatons or ncome generaton model taken from the nequalty decomposton lterature. For the purpose of ths paper, we gnore savngs and the nter-temporal and ntra-personal dstrbuton of welfare, thus assumng that ncome and expendture are equal as well as welfare derved from non-cash sources. Income Generaton Model The estmaton and smulaton of labor market partcpaton and earnngs s based on a structural model, where ndvduals partcpate n the labour market based on

5 ther own characterstcs and the decson of the household head, usually nfluenced by the hghest return on the actvty. The labour market partcpaton and household ncome model s nspred by the works n the development economcs lterature from an nter-temporal (Bourgugnon; Fourner and Gurgand 2001 and cross-country perspectve (Bourgugnon; Ferrera and Lete Table 2 descrbes a set of nested choce equaton descrbng the followng labour market characterstcs. They prmarly relate to labour market characterstcs and the presence of other market ncome sources. Table 2. Nested Labour Market Equatons In-Work Decson 2 Labor Force Partcpaton 3 In-work 1 Employed Self-Employed Farmer In-work 0 Presence of Captal Income Presence of Beneft Income Unemployed Retred Student Inactve In the model, we used the ndvdual as our basc unt of analyss, because each ndvdual belongs to a specfc household n each perod. In estmatng partcpaton n the labour market we used the bnary events such as n-work. We then draw a set of random numbers such that we predct the actual dependent varable n the base year. We estmate the economc status of the ndvdual. We establsh frst the economc status of the ndvdual by determnng whether or not the ndvdual partcpates n the labour market. The expected outcome from the logt estmaton we predcted on the probablty of whether each persons (mnmum 16 years of age partcpates or not n the labour market. Separate logt estmaton was used for both males and females. We defne our logt models as follows: y * Such that logt( p ln P o k 1 P B X k (1

6 y 1f y * 0 (2 In order to create the stochastc term,, we use the followng relatonshp: Such that A value of u that satsfes ths s: u ln (3 1 u k Bo X p 1 y 1 f u logt (4 k u Y r * p Y 0p r * p 1 1 (5 Where r s a unform random number. Once we have establshed, whether an ndvdual s n-work or not, ther work status, employed, self-employed and farmer. We also nclude f ndvdual has captal and beneft ncome. Once, we have establshed the labour force characterstcs of each ndvdual, ncome varables may be modelled usng an OLS equaton descrbed below: Y * exp X B (6 However gven the complexty of the system used n ths paper gven the requrements of smulatng earnngs, we choose a parametrc approach here along the lnes of the Bourgugnon; Fourner and Gurgand (2001 type analyss. Wthn our ncome generaton model, we nclude 5 market ncome sources; ncome from employment, self-employment, captal, benefts and remttances ncomes: Y M Y I Y I Y I Y I Y I (7 Emp Emp SE SE Farm Farm Captal Captal Benefts BeneftsR Where s ncome source and expressed as a functon of Y Y Y - f Z, Y t t,, t t, I I - g Z, I t t,, t t, I s the presence of ths ncome. Both can be

7 Y Where Z s a set of personal characterstcs, assocated wth the relevant model and functons. Y t t, and t t, and I t t, are the parameters I t t, the relevant dstrbutonal The equvalence scale for each household s calculated based upon the Organsaton for Economc Cooperaton and Development (OECD method. Ths apples a value of one to the frst adult and a value of 0.5 to all other adults and gves a value of 0.3 to each addtonal chld. Generalsed Entropy4 Generalsed Entropy can be a useful means of decomposng ncome nequalty. It can provde a better ft of the actual ncome dstrbuton and s consdered easy to decompose. The Half Squared Coeffcent s employed n ths paper (commonly referred to as the Generalsed Entropy Class of nequalty measures, I2. Ths places more weght on the ncome trend of hgh ncome households. The general formula of the Generalsed Entropy Class of nequalty measures s gven by GE 2 1 n 1 y 1 n 1 y (8 Where the y s are the ndvdual ncomes and s the arthmetc average n a populaton of n ndvduals. If everyone has the mean ncome, then the value s zero. The mean ncome dvdes the populaton nto an upper tal and a lower tal. In the upper tal the rato s above unty and t s below unty n the lower tal. If s equal to unty, then we have equal weghtng of the ratos. When s larger than unty, the hgh ncomes have even hgher ncome n the sum and the low-ncome ratos become even smaller n the measure of nequalty. When s smaller than unty, the lower tal ratos get closer to unty (become more mportant n the sum and the hgher value ncomes get pulled back to the mean. GE (2 s half of the square of the Coeffcent of Varaton y y y 1 1 y n 12 2 n. (9 1 One other measure that s wdely used n lookng at nequalty s the Lorenz curve based measure; the Gn coeffcent (fals decomposablty generally n n 1 Gn y yj (10 2n n 1 y j 1 1

8 One of the man axoms whch we usually requre nequalty measures to meet s that of decomposablty. Ths requres overall nequalty to be related consstently to consttuent parts of the dstrbuton, such as populaton sub-groups. For example f nequalty s seen to rse amongst each sub-group of the populaton then we would expect nequalty overall to also ncrease. Some measures, such as the Generalsed Entropy class of measures, are easly decomposed and nto ntutvely appealngly components of wthn-group nequalty and between-group nequalty. The Gn coeffcent fals decomposablty generally. Instead of usng a decomposton method for populaton groups, we can therefore use a decomposton method for ncome characterstcs. Inequalty s broken up nto the absolute factor contrbuton. Sf s defned n equaton (10. I f f S f II f (10 f f f f Where s the correlaton between component f and total ncome and f s factor f s factor share and s the GE (2 of total ncome multpled by GE (2 II f of the ncome source. The results n table 11 show the contrbuton of each ncome source towards nequalty n each regon for each household group. DATA AND TRANSFORMATION OF DATA In ths paper we use the 2004 Natonal Lvng Standard Survey data to examne the dstrbuton of ncome n Edo State. The Natonal Lvng Standard Survey s an annual survey of a sample of ncomes, labour market status and demographc nformaton collected by the Natonal Bureau of Statstcs (NBS (2004 Ngera. It s desgned to collect nformaton related to ncomes, expendtures, demographcs and labour market status as ts prmary objectve. One of the man objectves of the Natonal Lvng Standard Survey s to determne the poverty and nequalty stuaton n Ngera by measurng the level of the dstrbuton of resources among ndvduals and households. The sample sze s randomly selected from 120 housng unts from each state and the Federal Captal Terrtory (FCT, Abuja called Enumeraton Areas (EAs. The data set for our analyss comprses of a sample of ndvduals and households representng the total populaton that partcpated n the survey at any tme durng the perod. The concept of ncome used n ths study ncludes ncome earned both n cash and n-knd and s solely based on the determnaton of earnngs. The ncluson of weghts to account for non-response ensures that we can accurately represent changes n the actual populaton. We derved fve sources of ncome from the data. These are employee earnngs from wage work, non-farm self-employment earnngs, farm earnngs, captal earnngs derved from asset

9 dsposal/rental, and benefts derved from earnngs from old age penson and educatonal assstance (scholarshp. The economc status of the ndvdual s establshed.e. whether the ndvdual s employed, self-employed or a farmer. These questons can be answered by dentfyng whether or not the ndvdual s a recpent of earnngs from employment, self-employment or farmng. However, one of the lmtaton of usng a mcro survey data s that they are not updated perodcally especally n developng countres lke Ngera. For nstance the Natonal Lvng Standard Survey data 2004 used n ths study s the most recent mcro survey data n Ngera. The beneft of usng ths survey data s ther representatveness at the natonal level and the qualty of the data. Even though the data was survey 8 years ago, the results presented n ths paper stll represent the actual context n the composton of economc actvtes so far. If any changes n the compostons of the economc actvtes maybe nsgnfcant except for short term shock. Transformaton of the varables The transformaton of varables on educaton for nstance s transformed nto four broad levels of educaton.e. prmary, lower secondary, upper secondary and unversty. The age varable enables us to construct an age bracket that can be used to defne a chld. Ths s used to create varables for the number of chldren n each household by narrow age category e.g. zero years to fve, sx to ten years and eleven to seventeen years old. The martal status takes the form of sngle, monogamous marrage, polygamous marrage, separated, dvorce, wdow and nformal unon. We combne monogamous and polygamous marrage varables to represent all those ndvduals who are currently marred. Summary statstcs The followng tables 3 to 6 show the demographc varables used for the decomposton method. These demographc varables nclude age, household sze, educaton levels and employment partcpaton. These offer some clues as to the lkely source of changes n ncome nequalty among the regons. Table 3: Average Age of Household Heads and Number of Persons per Household Household groupngs 5 Average age of household heads Number of persons per households

10 Total 47 4 Source: Estmaton by the author s As can be seen from Table 3 ages of the household heads were estmated for all the household groups n Edo State. It s evdent that average age of household heads for groups 2 and 11 appears not to change much. However, the average age n all other household groups appears not sgnfcantly dfferent from each other. Nevertheless, household group 6, 7 and 10 has the average household sze of 5, 6 and 7 respectvely. Ths shows that these groups are marred couples wth chldren and other adults are lvng together. Table 4: Percentage of Adults, Chldren and Persons over 65 years old by Household groupngs Household Adults Chld 8 groupngs Total Source: Estmaton by the author s The percentage of adults aged years old n the populaton appears to be hgher n Edo State as shown n Table 4. The proporton of chldren account for 36.4 percent and the older populaton n Edo State.e. 65 years and above consttute 4.7 percent respectvely. However, most household groups contanng chldren follow the trend of a reducton n the proporton of chldren. Household groups 1, 2, 4, 7, 9 and 11 do not nclude chldren. The reducton n the proporton

11 of chldren s lkely to have ncreased the equvalsed ncome of chld-rearng households and also lkely to have reduced nequalty. Decson regardng labour supply s lkely to have nteracted wth decson regardng famly sze and these decsons may appear to have worked towards reducng nequalty.

12 Table 5: Percentage of Adults aged 16 and above In-work, Employed, Selfemployment, Farmer, Captal and Beneft Income Recpent by Household groupngs Household groupngs 9 Inwork Employed Self employed Farmers Hascaptal Hasbeneft Total Source: Estmaton by the author s Lookng at the proporton of adults who are workng, table 5 was constructed from addng the proporton of employed, self-employed and farmng actvtes. The result reveals that the proporton of adults aged 16 years and above s 72.5 percent n Edo State. From ths total, farmng actvtes accounted for hghest proporton of adults (34.3 percent. Wage employment followed wth 22.7 percent. We also observed from Table that the proporton of adults engaged n farmng actvtes s the hghest for sngle old persons (Group 2 wth 80 percent. For wage work, sngle persons (Group 1 form the hghest category wth 41.6 percent and followed by marred couples wth more than two adults wth chldren accountng for 24.3 percent (Group 6. Ths sgnfes that sngle household heads partcpate more n the labour market n Edo State than self-employed household heads. The proporton of household heads wth captal and beneft ncome vares between household groups wth household (group 10 and household (group 7 accountng for 25.7 percent and 26 percent respectvely. Educaton The mportance of educaton n determnng the ncome of an ndvdual, household or country s well documented. Table 6 shows the mprovement n educaton level of household heads n Edo State. It s mportant to note that one must acheve a degree n order to be classfed as havng unversty educaton. Dplomas, HSC certfcates and other post leavng certfcate achevements are consdered as

13 beng upper secondary, whle secondary educaton such as Junor and Senor Secondary School are consdered as beng lower secondary educaton. Ths dstncton s made so that one can determne the mpact of havng a unversty degree. Table 6: Partcpaton Rate of Household head by Educatonal attanment (for household heads Educatonal attanment Inwork Employed Selfemployed Farmer Hascaptal Receved benefts No educaton Prmary Lower 0 8 secondary Upper secondary Unversty Total Source: Estmaton by the author s Table 6 shows the percentage of household heads by educatonal level n each employment status. The proporton of unversty educated earnng employment ncome s 76 percentages, whle 72 percent of self-employed ncome comes from those households wth prmary educaton. The partcpaton of household heads n the employment sector s hgher for unversty educated and ths reterates the mportance of educaton n determnng employment partcpaton especally n the formal sector. However, one should consder that elderly people are less lkely to have upper secondary or unversty educaton as reflected n the column for the self-employment households. So the results are by no means a conclusve assessment of the relatonshp between educaton and employment partcpaton rate. The majorty of household heads n households wth farm ncome had nether a prmary leavng certfcate nor thrd level qualfcaton. The hgh rate of employment partcpaton s therefore accompaned by the hgh rate n the educatonal level of many households. Ths underlnes the mportance of educaton n mprovng the employment prospect for many ndvduals. The ncome dfferental between unversty and non-unversty graduates s such that an ncrease n the number of unversty graduates s lkely to have ncreased nequalty especally gven the low ntal base.

14 REGRESSION RESULTS: OCCUPATIONAL CHOICE AND LABOR MARKET INCOME The model s estmated separately for both males and females who partcpated n the labour market. We estmated for those who are workng (n work, employed, self-employed, farmers, and recpents of captal and benefts ncome on demographc characterstcs. Table 7 shows the results of the logt method we used for the estmaton. We ncluded dummy varables that capture the maxmum educatonal level acheved. The omtted category s no educaton. A martal status dummy varable, age and age squared were also ncluded n the regresson. The dummy varable takes the value 1 when the ndvdual obtaned unversty or upper secondary school certfcate. Followng Bourgugnon; Ferrera and Lete (2002, our analyss assumed that labour market partcpaton choces were made wthn the household as mentoned earler. The coeffcent of most educatonal level are postve, sgnfcant, and ncreasng wth the educatonal level, that s, the returns to educaton are always postve especally for males. The most strkng features emergng from the estmatons s the hgher partcpaton rate of females n wage employment than males n the unversty and upper secondary levels as shown n Table 7. These varables are all statstcally sgnfcant at 1 percent level. In addton, partcpaton rate ncreases wth marrage and decreases wth rural areas and number of chldren. For the self-employed sector t decreases wth marrage and rural areas. Farmers partcpaton n the labour market decreases wth educaton and employment as expected and ncreases wth rural areas and marrage. Beneft recpents ncrease wth educaton and rural areas and decreases wth age.

15 Table 7: Labour Market Partcpaton Equaton usng Logt method for Inwork, Employed, Selfemployed, Farmer, Has captal, Receved Beneft Income for both Males and Females Source: Estmaton by the author's Estmaton of the labour market earnngs of household members In ths sub-secton, we estmated for three man sources of households ncome: employed ncome, self-employed ncome (non-farm and farm ncome. Others sources added nclude captal ncome and benefts ncome. For employed or wage ncome, a logarthm of the yearly wage s estmated usng the Ordnary Least Square (OLS method. In the estmaton procedure, wages were mputed for ndvduals workng as employed workers, but for those whom

16 earnngs had not been observed we used the estmated wage equaton. The estmaton s specfed separately for males and females and a dummy varable for rural locaton s ntroduced. To account for the multple actvtes of some employed workers, a dummy varable s ntroduced n the wage equaton takng the value of 1 f the ndvdual supples labour as employed wage earner. The dependent varable of the average earnngs from weeks worked were derved from dvdng the earnngs from the number of weeks worked. Weeks wth work nclude all persons of ether sex wthn the specfed age lmt (16 years and above who furnsh the supply of labour for the producton of goods and servces durng a specfed tme reference perod. The Ordnary Least Square (OLS method was used for the estmaton of the yearly average wage earned. However, the estmated coeffcents of the wage equatons n Edo State show a general decrease n return to educaton of wage earnngs and ncrease n return to educaton of self-employed earnngs wth a decrease n wth experence. Ths result s n lne wth the theoretcal expectaton that returns to educaton ncrease for each supplementary year. Wage earnngs are not n consensus to ths asserton, however, the returns to prmary and lower secondary educaton n Ngera has fallen substantally snce 1966 as reported by Psacharopoulos (1985. Ths estmaton seems to be n lne wth these assertons. The coeffcents age and age squared n the wage equaton suggest an nverted U-shaped wage-age profle. The most strkng feature from the estmaton of farm ncome s that t shows a declne n the rural areas. We also observe a reducton of farm ncome among the educated people. For captal earnngs, the estmaton shows a decrease among males and females wth educaton and nwork. Benefts ncrease wth educaton for both males and females decreases wth lower educaton for males. The mpact of beneft n the rural areas also reduces for both males and nwork for males and females as expected.

17 Table 8: Choce of Occupaton and Labour Market Income (OLS Estmaton Employment ncome Indepenent Males varables Unversty * * (2.555 Upper - secondary 3.556* * (1.823 Lower secondary * * (1.421 Prmary (1.343 Rural (0.321 Marred O.488 (0.455 Age (0.177 Age (0.002 Experence * * (0.177 Female s ( ( ( ( ( ( ( (0.160 Selfemployment ncome (nonfarm Males Female 2.829* * ( * * ( ( ( (0.067 s ( ( ( ( ( ** * (0.027 Farm ncom e Captal ncome Males Males Female s (0.403 (1.870 ( ( ( ( ( ( ( ( ( ( * * ( * * ( ( ( ( ( ( (0.000 Benefts Males ( ( ( ( ( ( (0.001 Female s 6.957* ( * ( ** ( ** ( ( ( (0.001 Experence ( ( ( ** ( (0.000 Inwork

18 Constant * ( ( * ( * ( * (0.584 ( (3.767 (1.130 ( * (3.749 * (1.760 ( (1.539 R observato ns Source: Estmaton by the author s RESULTS: PRELIMINARY EXAMINATION OF TRENDS IN INEQUALITY Ths secton uses the Natonal Lvng Standard Survey (NLSS 2004 to analyse the dstrbuton of ncome n Edo State. Perhaps the best startng pont s to see how varaton between household groups allows one to dentfy the contrbuton of demographc change towards nequalty. The result n table 8 shows the conjectured amounts from the sample populaton on the average household ncome and ncome by source for each household group. One can see from the followng tables the household groups wth the lowest and hghest ncomes n Edo State.

19 Table 9: Equvalsed Market Income and Average Nomnal Income by Source and Household Group (ncludes all Households for all ncome sources Household Groups Total Total market ncome Employed ncome Selfemployed ncome Farm ncome Captal ncome Benefts Source: Estmaton by the author's

20 As we can see the average equvalsed market ncome vares among the household groups. Marred household wth chldren and other adults n the households.e. (group 6 seems to have the hghest total market ncome, followed by sngle household wth chldren. The calculatons nclude those households wth zero values. Most household groups have hgh employee ncome. Employee ncome vares between the household groupngs. The average employee ncomes tend to be hgher among sngle household heads than other household groupngs n Edo State. Self-employment ncome s hgher n household group 5, 6, 7. These households have more than two adults and chldren n ther households. The reason mght be that there are more than two adults lvng n these households and workng. Whle the average farm ncome s partcularly hgher than other sources of ncome. We equally observed that household group 6 and 7 have a hgher ncome from farm work than other households. The mplcaton of ths s that these households have large number persons that partcpate n farm work. Ther partcpaton mght have led to hgher ncome wthn the household. Captal and benefts earnngs are hgher among marred households wth chldren.e. group 5, 6, 7 and 8 respectvely.

21 Table 10: Inequalty wthn and between Household Groups n Edo State Household Pop share Rel. Gn 12 groupngs 9 Mean Source: Estmaton by the author s Table 10 shows that wthn-group nequalty was greatest for household groups 1, 3, and 10 and lowest for group 2, 7 and 9. However, the ncrease n nequalty wthn household group vared between household groups. For nstance, the hgh rate n wthn-group nequalty s greatest for groups 1, 3 and 10. The most unequal household groups 1, 3, 4 and 10 are hgher n terms of populaton share. The collectve share for the least unequal groups 1, 6 and 10 remaned unchanged wthn households. These trends further support the case that demographc trends ncrease nequalty wthn households. Table 11: GE (2 and Absolute Factor Contrbuton of each Income Source (Decomposton Results by household groups n Edo State Househ old groupn gs 13 GE(2 Employed ncome Selfemployed ncome Farm ncome Captal ncome Benef ts Total

22 In table 11, the contrbuton of all the determnstc factors to nequalty n Edo State, based on the GE (2 decomposton, usng the regresson based decomposton method s presented. The decomposton of contrbuton to ncome nequalty by the resdual term, the constant term and all non-constant Xs are carred out for each ncome source and for each household group. From the result we observed n household group 7 for example that the GE (2 value of 354 s hgher relatve to that of other household groups. Household group 4 has a GE (2 value of 65. Ths suggests that nequalty wthn group s much hgher for marred couples wth more than two adults but wth no chldren under the age of 65 (group 7 than for all of the other groups. Farm ncome appears to be the bggest contrbutor towards ths nequalty wthn the group for marred couples wth more than two adults but wth no chldren. The absolute factor contrbuton s +349 for farm ncome and -3 for self-employment ncome and -1 for benefts. Ths shows that self-employment and beneft ncome work towards reducng nequalty wthn the group. The mportance of captal and benefts towards overall nequalty s nsgnfcant n most household groups n Edo State. However, benefts tend to reduce nequalty, but n ths case t has lttle or no mpact. Table 12: Inequalty of Equvalsed Household Dsposable Income by Age Group (All Households n Edo State Age Pop. share Relatve mean Gn group Less than Above Source: Estmaton by the author s The statstcs accordng to household group appear to suggest an ncreasng nequalty from demographc change. Some of those household groups are defned accordng to the age of the household members. The results presented n table 11 specfcally show the extent of wthn group nequalty for dfferent age group n Edo State. Wthn group nequalty s hgher for most age groups. In Edo State, wthn group nequalty was greatest for households headed by ndvduals between less than 25 and above 65 years old and lowest for households headed by ndvdual between the ages of 45 and 54 years. The hgh nequalty wthn the less than 25 and 44 years old bracket s a reflecton of long term unemployment among some

23 persons n that age bracket. The mpact of ncome changes between age group s lkely to have caused the hgh rate of nequalty. Table 13: Gn Coeffcent for each Component of Household Market Income Decomposton due to demographc change Employed Selfemployed Farm Captal Beneft Total Gn Source: Estmaton by the author s The combnaton of employed, self-employed, farm, captal and benefts ncome forms market ncome. It s mportant to assess the mpact of demographc change on ncome nequalty. Table 13 shows the mpact of demographc change upon the nequalty on equvalsed market ncomes n Edo State. The result shows that the levels of nequalty dffer slghtly for each component of household ncome. As we can see captal and beneft ncome account for the hghest level of nequalty of and respectvely. Ths mples a slght unequal dstrbuton of ncome for each ncome components. These decomposton results are consstent wth the pattern antcpated on the bass of the data. CONCLUSION AND SUMMARY Ths study takes a mcroeconomc approach to assess ncome nequalty n Edo State. The analyses dentfy the sources of nequalty and ts demographc patterns. Specfcally, our framework allow us to account for the mportance of ndvdual decson such as labour partcpaton and household structure, whle at the same tme presentng nformaton about the mportance of dfferent ncome sources. We use a number of measurements of nequalty to measure the level of nequalty at dfferent ponts of the dstrbuton of ncome n Edo State. However, our results can only be nterpreted as uncoverng the causes of nequalty under specal crcumstances, so t s probably safer to vew ths as an accountng procedure that dentfes the sources of nequalty from a statstcal pont of vew. In any case our results do reveal that educaton plays an mportant role n the partcpaton of labour market. In terms of average household earnngs they are relatvely hgher among households wth wage employment as expected. In regard to the decomposton, we found an ncrease n earnngs due to demographc change wthn and between household groupngs. Age seems to have resulted to hgh rate of nequalty due to ncome change between age groups. Wthn group nequalty was hghest for households headed by ndvduals between less than 25 and above 65 years and lowest for households headed by ndvdual between the ages of 45 and 54 years old. The gap between households headed by such adults and other households appears to be hgher. Ths seems to have contrbuted towards hgher nequalty.

24 Usng the regresson base decomposton method to decompose ncome nequalty nto contrbutons by ndvdual determnstc factors plus the shares by the resdual term. We found that base on the GE (2 decomposton, farm ncome s the greatest factor towards overall nequalty for most household group. The mpact of self-employment, benefts works towards reducng nequalty wthn the group.

25 ENDNOTES 1. For a detaled framework of the method used see Bourgugnon, Ferrera, Lete ( Explanatory Varables (Labour Market: Unversty educated, Upper secondary educated, Lower secondary educated, Prmary educated, Marred, Number of chldren, Age, Age 2, wdowhood, Locaton Type (rural, experence, experence2 3. Explanatory Varables (Earnngs: Unversty educated, Upper secondary educated, Marred, Lower secondary educated, Prmary educated, Experence, Experence2, worky, wdowhood, Number of chldren, Age, Age 2, Locaton Type (rural. 4. Entropy s a measure for how many transformatons already have occurred n a system. If completely equal dstrbuton (of whatsoever n a system leads to maxmum entropy of that system and f low entropy of that s caused by hgh dstrbutonal nequalty, then achevng equal dstrbutons means that the dstrbuton process s saturated = sngle person wthout chldren, 2 = sngle old person, 3 = sngle wth chldren, 4 = marred couple no chldren, 5 = marred couples wth chldren, 6 = marred more than two adults wth chldren, 7 = marred more than two adults wth no chldren, 8 = unmarred adults wth chldren, 9 = unmarred wth no chldren, 10 = marred more than two adults more than two chldren, 11 = marred old couple 6. Adults refer to ndvdual aged years old refer to adults 65 years and above 8. Chld s under 16 years old 9. 1 = sngle person wthout chldren, 2 = sngle old person, 3 = sngle wth chldren, 4 = marred couple no chldren, 5 = marred couples wth chldren, 6 = marred more than two adults wth chldren, 7 = marred more than two adults wth no chldren, 8 = unmarred adults wth chldren, 9 = unmarred wth no chldren, 10 = marred more than two adults more than two chldren, 11 = marred old couple = sngle person wthout chldren, 2 = sngle old person, 3 = sngle wth chldren, 4 = marred couple no chldren, 5 = marred couples wth chldren, 6 = marred more than two adults wth chldren, 7 = marred more than two adults wth no chldren, 8 = unmarred adults wth chldren, 9 = unmarred wth no chldren, 10 = marred more than two adults more than two chldren, 11 = marred old couple

26 11. Relatve mean-equvalsed ncome relatve to the average household 12. Gn of Equvalsed Household market ncome for each household group = sngle person wthout chldren, 2 = sngle old person, 3 = sngle wth chldren, 4 = marred couple no chldren, 5 = marred couples wth chldren, 6 = marred more than two adults wth chldren, 7 = marred more than two adults wth no chldren, 8 = unmarred adults wth chldren, 9 = unmarred wth no chldren, 10 = marred more than two adults more than two chldren, 11 = marred old couple 14. Ths result s n lne wth the earler fndngs of Oyekale et al, (2006, who noted that employment ncome ncreases nequalty whle agrcultural ncome decreases nequalty. we also use the Gn ndex n the dscusson n ths secton

27 REFERENCES Adams, R.H. and J. Jane (1995, Sources of Income Inequalty and Poverty n Rural Pakstan, Research Report 102 Int. Food Polcy Research Insttute 17(2 80 ISBN: Addson and Corna (2001. Income Dstrbuton Polces for faster poverty Reducton. WIDER Dscusson Paper No. 2001/93, World Insttute for Development Economc Research. Awoyem, T.T. and A.I. Adeot (2004, The Decomposton of Income Inequalty by Sources of Income: The Rural Ngeran Experence. Afrcan Journal of Economc Polcy 11 (1: 1-16 Barro, R (2001. Inequalty, Growth and Investment, n K.A. Hasselt and R.G. Hubbard, eds; Inequalty and Tax Polcy, AEI Press. Brdsall, N. Ross, D. and Sabot, R. (1995. Inequalty and Growth Reconsdered, World Bank Economc Revew, Vol. 9 (September, pp , Brdsall, N. and Juan, L. (1997. Asset Inequalty Matters: An Assessment of the World Bank s Approach to Poverty Reducton. Amercan Economc Revew, Vol. 87 (May. Bhalla, S. (2003. Recountng the Poor: Poverty n Inda, Economc and Poltcal weekly, January: Blnder, A.S. (1973: Wage Dscrmnaton: Reduced Form and Structural Estmates, Journal of Human Resources, 8, pp Bourgugnon, F.H. G. Ferrera and P. G. Lete (2002: Beyond Oaxaca-Blnder: Accountng for Dfferences n Household Income Dstrbuton Across Countres, Journal of Economc Inequalty 6, pp Bourgugnon, Franços, Martn Fourner, and Marc Gurgand Fast Development wth a Stable Income Dstrbuton: Tawan, Revew of Income and Wealth 47(2: Clarke, G., L. Coln, XH. Zou, (2003. Fnance and Income Inequalty: Test of Alternatve Theores. World Bank Polcy Research Workng Paper 2984, Washngton DC. World Bank. Escobal, J. and M. Torero (2005 Adverse Geography and dfferences n welfare n Peru In Rav Kanbur and Anthony Venable (eds, Spatal Inequalty and Development. Oxford Unversty Press. January.

28 Fredman, J. (2005 How Responsve s Poverty to Growth? A Regonal Analyss of poverty, Inequalty and Growth n Indonesa, In Rav Kanbur and Anthony Venable (eds, Spatal Inequalty and Development. Oxford Unversty Press. January. Frawley, J., P. Commns, S. Scott and F. Trace, (2000. Low Income Farm Households: Incdence, Characterstcs and Polces, Dubln: Oak Tree Press Kanbur, R. and N. Lustg (1999. Why s Inequalty Back on the Agenda? Paper prepared for the Annual Bank Conference on Development Economcs, World Bank Washngton DC Aprl NBS (2004, Ngera Lvng Standard Survey, Natonal Bureau of Statstcs, Statstcs Dvson, Government of Ngera, Natonal populaton commsson (2006 Populaton Census Estmates Oaxaca, R. (1973: Male-Female Wage Dfferentals n Urban Labour Markets, Internatonal Economc Revew, 14, pp Oyekale, A.S, A.I., Adeot and T.O. Oyekale, (2006 Measurement of Sources of Income Inequalty n Rural and Urban Household n Ngera. Poverty and Economc Polcy Network Workng Paper, Unversty laval, Canada pp Psacharopoulos, G. (1985, Returns to Educaton: Further Internatonal Updates and Implcatons, Journal of Human Resources, Vol. 20, No.4 pp Rewane, B. (2011 Income Dsparty between North and South wdens. Ths Day Newspaper 15 May 2011 Sala--Martn, X. and Subramanan, A. (2003 Addressng the Natural Resource Curse: An Illustraton from Ngera, IMF Workng Paper WP/03/159 Sahn, D and D. Stfel (2003. Urban-Rural Inequalty n Lvng Standard n Afrca. Journal of Afrca Economcs Vol. 12, Number 1 December, pp