By Nguli,Ndunge FEMALE PARTICIPATION IN LABOUR FORCE IN GREATER NAIROBI.

Transcription

1 FEMALE PARTICIPATION IN LABOUR FORCE IN GREATER NAIROBI. I By Nguli,Ndunge UNIVERSITY OF NAIROBI E' -^fucana COLLECTION Research paper submitted to the Department o f Economics, University o f Nairobi in partial fulfillment for the award of the Degree of Master of Arts in Economics. September, uat/o '1<1NYA7TA MEMORIAL LIBRARY

2 This Research Paper is my original work and has not been presented for a degree in any other university. M y Ngulj,tfdunge. This Research Paper has been submitted for examination with our approval as University Supervisors. Dr. Damiano Kulundu Manda. Mr. Patrick Machyo.

3 T A B L E O F C O N TE N TS a c k n o w l e d g e m e n t i A B S T R A C T......ii TA B LE O F C O N T E N T S... iii LIST O F T A B L E S iv C H APTER O N E Introduction*..., S tatem ent o f the P roblem......' ^..O bjectives o f the S tudy...! Justification o f the S tudy... 5 C H APTER T W O...;...' LITER A T U R E R E V IE W Em pirical literature re vie w...6 ; 2.1.1R eview o f Individual S tudies ve rv ie w o f Literature...17 C H APTER M E T H O D O LO G Y S am pling procedure., ^ M odel sp ecifica tio n......" :3.E stim ation te c h n iq u e T heoretical fra m ew o rk C H APTER 4... ; D escriptive S tatistics T.O. C haracteristics o f the sam ple Estim ated M odel R esults C H APTER F IV E C O N C LU S IO N AND PO LIC Y IM P LIC A T IO N S C onclu sion Policy im plications R E F E R E N C E S in

4 LIST OF TABLES Table 1: P articipation and unem ploym ent rates for rural and urban a re a s...2 Table 2: S am pling procedure...20 T a ble 3: D istribution o f the population by G ender G e neration Table 4.1: Fem ale Participation in the labour fo rc e...28 Table 4.2: Fem ale Labour force participation by a g e Table 4.3: Fem ale Labour Force participation and N on-labour incom e Table 4.4:F em ale Labour Participation and E d u c a tio n Table 4.5: M igration Status and Labour P articipation...30 Table 4.6: Labour Force Participation by religion Table 4.7: Labour P articipation arid M arital S tatu s Table 4.8 : C orrelation m atrix for all the varia bles Table 4.9: R egression re s u lts...34 l I l iv

5 I A C K N O W LE D G E M E N T I am greatly indebted to my supervisors, Dr. D am iano Kulundu M anda and Mr. P a trick'm a ch yo fo r their invaluable assistance and guidance throughout this paper. T hey read all the drafts of this research paper and m ade constructive criticism s. I w ish to thank the U niversity o f Nairobi fo r granting me a scholarship #to pursue this course. I cannot forget to thank Prof. G erm ano Mwabu, a man of G od w ith a kind heart. He w as alw ays there to assist w ith his w ide know ledge. I th a n k him also fo r the.keen interest he had in this stu dy and fo r the useful i i comments he gave. God bless you Professor, v t I * ' D uring th e course, I received assistance from m y co lle ag ue s and especially Bernadette Wanjala and Jack Jones. I thank them all for their valuable comments. I would also like to thank Reuben, Njambi and Beldine for their moral support, M y brothers and sisters, they are m any but I list them all: - Charles, Peter, Maurice, Angela, Ann, Felista and Marrieta. All of you are precious just like i diam ond. T hank you for your prayers and encouragem ent. I also thank my parents,.m ichael and Bernadette Nguli fo r their love and support, and also for j ' f ed ucating me. - i

6 i ABSTRACT There are several factors that affect the participation of females in the labour force. The paper focused on identification and analysis of factors that determine female labour force participation in Greater Nairobi. A logit model was estimated. The study found that only 34 percent of the women interviewed were in labour force. Factors such as age of the woman, - marital status, migration status, education attainment, religion and experience had a significant impact on female labour participation. l 11

7 , ' C H A P TER O NE 1.1. Introduction» The labour m arket has' a m ajor role to play in the econom ies o f developing countries. First it is an im portant channel fo r the transm ission o f both external disturbances and adjustm ent policies. For instance, labour m arket flexibility is im portant in reducing unem ploym ent endured in the adjustm ent process. Secondly, labour m arkets in developing countries play an im portant role in determ ining the level arid distribution of income. Individuals w ith regular w age em ploym ent are likely to be in the m iddle and upper incom e groups, w hile those w ithout em ploym ent are am ong the poor. How ever, our understanding o f labour m arkets in developing countries is still far from adequate (IVIanda 1997). In Kenya, there are large differences betw een the urban participation rates for men and wom en. The fem ale urban participation rate increased from 38.8 percent in 1977/78 to 56 percent in D espite this increase, fem ale participation rate w as I. 26 percentage points low er than m ale urban participation rate (82%) in In the rural areas, the participation rates for both m ales and fem ales are very high and the difference between them is.small (see Table 1). For instance, in 1989 the male participation rate in the rural areas w as 86.9 percent and the fem ale participation rate w as 91%. \

8 Table 1: Participation and unemployment rates for rural and urban areas (percent). Urban male Urban female Rural male Rural fennale Y ear' 1977/8 1, / /8 1988/9 1977/8 1988/9 Participation Unemployment rates Source: " 0.1 Republic o f Kenya (1988' and W ilson (1991) Table Labour force Survey R eport 19 36; Cam er 6. T h e urban participation rates increase w ith the level o f com pleted education, particularly for fem ales. Participation rates fo r U niversity trained w om en range between 20 and 30 percent higher than the average participation rate fo r all wom en (Milne, and Neitzert, 1994). However, for other levels o f education, the low participation rate m ay be attributed to low education attainm ent am ong wom en, fam ily obligations and inflexible w prking tim e especially in w age em ploym ent. Unemployment is conventionally defined to include persons of working age seeking work at going wages, but unable to find any. M ajor conceptual and m easurem ent difficulties m ake the estim ates o f unem ploym ent in an econom y, such as Kenya, to be.imprecise. U nem ploym ent figures for the entire econom y are very rare, but estimates for rural and urban unemployment based on labour force surveys are available. U nem ploym ent estim ates based on the rural labour force survey 1988/89 shows that rural unem ploym ent in Kenya w as 0.3 percent, w hich is very low com pared w ith urban unem ploym ent estim ates (37.7 percent). Urban unem ploym ent seem s to have rem ained the same from the late 1970s to the m id 1980s. For instance, the estim ated un em p lo ym e nt rate w as 16 percent in 2

9 t 1986, which was about the same as in This shows that during the nine year's, the.econom y created jobs that absorbed the increasing labour force over the period. However, recent estim ates show that the urban unem ploym ent situation in Kenya.has deteriorated since then. Unem ploym ent estim ates for the early 1990's show that urban unem ploym ent ranged between 7.8 and 23.6 percent (R epublic of Kenya, 1994, and W orld Bank, 1993). The high unem ploym ent in th e 1990s can be.attributed to the governm ents restrictive policy on em ploym ent in the public sector and the civil service reform s com bined w ith the restructuring program s o f m any private se cto r large em ployers to cope w ith the em erging co m pe titive business \ environm ent. Evidence by sex, show that unem ploym ent am ong fem ale m em bers o f the labour force w as m ore than double (24.1% ) that for m ales (11.76) in 1986 (R epublic of Kenya, 1988 and 1990). Also, unem ploym ent m ainly affects the young, w ith those aged 15 to 30 representing 80 percent of the unem ployed. This supports the ILO (1972) conclusion that unemployment burden falls heavily on wom en and the youth. M ilne and W ilson (1991) attribute the high unem ploym ent am ong young participants to the fact that new entrants to the labour m arket take tim e to adjust their expectations to the type o f jobs available. Evidence by education attainm ent shows that unemployment was highest among those with secondary school education. This is consistent w ith the evidence by collier and Lai (1986) o f long delays between leaving school and starting w ork for school leavers. Sim ilarly, U niversity graduates have a high unem ploym ent rate pa rtly due to the policies ad op ted by the 3

10 governm ent fo r exam ple, the civil service dow nsizing. The governm ent no longer guarantees em p loym e nt fo r all U niversity graduates and it is likely that the unem ploym ent rate fo r U niversity graduates will continue to increase. \ * 1.2. Statement of the Problem, A vailable literature show s that w om en have low er labour participation rates and higher unem ploym ent rates than their m ale counterparts (especially in urban.areas). In 1986, ju st over two thirds of the total urban w orking age population participated in the labour force, 56 percent o f all w om en and 82 percent of all men (R epublic of Kenya 1988). In this spate of gender reverberations, it is im perative to understand the factors that w ould lead to the decision by Kenyan fem ales to enter the la b o u r m arket. Like in other de ve lopin g countries, participation of w om en in ' ' i developm ent and in incom e generating activities is restricted by m any constraints such as childcare, household responsibilities, lack o f education, lack o f experience and so on. M oreover, clear-cut guidelines about integrating w om en in developm ent process and defining their roles are still lacking. Policy m akers generally feel the shortage of reliable information on the current socio- econom ic and dem ographic status o f w om en in developing countries. D espite this observation, only a few studies have been done to elucidate the factors responsible fo r w om en participation in the labour force so that policy m akers can.incorporate them in decision making. Some studies like M aglad s (1998), study fem ale labour supply using m arried w oipen only. T h is is considered inappropriate fo r fem a le lab ou r su pp ly studies in i developing countries. Furtherm ore, reconnaissance survey in Nigeria show that a significant pe rcen ta ge o f fem ale labour participants are not m arried (single, 4

11 5 divorced, or separated). This is a sam pling bias problem that the current study seeks to address Objectives of the Study The broad objective o f the study is to analyze the determ inants o f fem ale labour force participation in G reater Nairobi. Specific objectives are: - 1) Identify the factors affecting fem ale la b o u rfo rce participation. 2) A n a lyze the facto rs determ ining fem ale labour force participation in G reater Nairobi. i > ' 3) S uggest policies to enhance and im prove w o m e n s hum an resource developm ent. ' Justification of the Study In the recent years, governm ents of developing countries including the Kenyan government have begun to consider economic development or. w hich w o m e n s i hum an resource developm ent is identified as a key priority. However, there is a dearth o f system atic studies o f determ inants o f w om en participation in the labour force, m aking it d iffic u lt'fo r policy m akers to com e up w ith policies to prom ote women participation in nation building. O ur study w ill provide policy m akers with indicators that can readily be used to improve women participation in the labour force. CHAPTER TWO

12 2.0. LITERATURE REVIEW 2.1: Empirical literature review T he fo llo w in g se ction review s em pirical studies on the determ inants o f w om en \ ' i ' participation in the labour force in various parts of the world. An attem pt is made to contrast and com pare the em pirical results w ith the approach o f this study Review of Individual Studies D em ographic and social barriers affect w om en s participation in the labour force. Negatu (1993) argues that differences in labour supply behaviour usually arise from disparities in productivity endowm ents, including dem ographic variables such as age, sex, and m arital status. M aglad (1998) in his study o f Sudan attem pts to capture the dynam ics o f fem ale labour supply in the m odern sectors. He used a sub t _ sam ple o f m arried w om en resident in Urban Sudan as the basis for his analysis. From his findings childcare responsibilities are argued to have negative im pact on w om en's m arket participation. * i M anda (1997) argues that education is m ore im portant in influencing fem ale than male participation decisions. C ollier (1990) argues that once in the labour market, wom en earn equal pay to that o f men, controlling for their characteristics. N evertheless w om en are less likely than men w ith sim ilar characteristics to enter the lab ou r m arket but g e n d e r differences In participation n a rrow as education # ' * i increases. In a study for Indonesia, D eolalikar (1992) argues that m ales earn significantly m ore and participate m ore in the labour m arket than fem ales at all levels due to average differences in levels o f schooling. 6

13 Job tenure and experience also influence labour participation and the gender w age gap. A ppleton et al (1999) argues that lack o f experience and discrim ination against m arried.women is a plausible explanation for greater gender differential. Behrm an and W olfe (1984) also argue that experience play substantial roles in determ ining labour force pa rticipation and earnings, as w ell as in sorting am ong sectoral labour 4. force participation. M eng and M iller (1995) also find that job tenure has a strong :and positive im pact on earnings in aggregate w hile jo b experience has a m oderate positive effect on earnings. N egatu (1993) supports these studies and argues that \ ' experience and the nature o f the labour m arket itself leads to differences in labour C / m arket participation by gender. Dabalen (2000) shows that in Kenya w om en with the, least, skills saw th e ir position w orsen relative to m en w ith sim ilar skills, even as r * women with the most skills were gaining ground on comparable men. A nker and Know les (1978) conducted a survey on the determ inants o f fem ale labour force In Kenya using data from tha N ational H ousehold S urvey in T he total sam ple available fo r analysis consisted o f 3,180 households, 701 o f w hich (22 1 per cent) were enumerated in urban areas. A nalysis w a s conducted in two stages: (a) On how micro variables affect female labour force participation rates, and (b) On how m acro-level variables describe labour m arket conditions in each sam ple town. 7

14 M acro level variables were used together w ith m icro variables. O rdinary least- squares m ethod of estim ation was used. The result of their study w as as follows: - The m icro-level variables that were found to be significantly affecting fem ale labour force participation w ere -Education of fem ale (secondary school level), Income (household incom e) and m arital status.. The m icro-level variables that w ere statistically insignificant were: -Child care re la te d variables and the practice o f Islam. T hey concluded that jobs in Kenya urban areas are m ost likely to be obtained by w om en w ith better education and by wom en w ith better connections as reflected by the fam ily incom e variable. The M acro-level variables used for the analysis were: - Health, U nem ploym ent rate proxy (percent of adult population not employed in the modern, non-agricultural sector), Average modern sector wage in service sector (in thousands of Kshs), and S ervice (percent o f non-agricultural m odern sector em ploym ent in services.) W om en living In tow ns w ith a relatively unfavourable jo b m arket tend to have low er labour force participation rates. W om en living in tow ns w ith a relatively favourable Job market in the Industrial sector most relevant to them i.e. the service sector, tend to have hig h e r la b o u r participation rates. T h e y concluded that be tte r educated,» single w om en from high-incom e fam ilies w ere most likely to be in the labour force in urban areas. 8

15 f Lack o f assets not only leads to lower participation by wom en but also constraint girls, access to education. Alderm an and King (1998) indicate that the absence of cash earnings in m any societies lim its the ability of w om en to realize and remit m arket returns from their education and thus reduces the signals to girls and parents about the desirability o f girl s education. These argum ents supports A ppleton et al (1990) w ho argue that asset incom es have a negative im pact on. w ork decisions and participation rates. Bigsten and Horton (1997) also argue that b ia se s w ithin the fam ily affect the am ount of hum an capital w om en acquire and that girls get less education because parents think the benefits accruing to sons will be higher and thus m ay have pro-son bias. Neitzert (1994) argues that wom en participation in the paid labour m arket is curtailed relative to their m ale counterparts beca use.the labour m arket provides incentives w hich tend to reproduce existing sexual division, of labour in w hich wom en specialize in household and subsistence prediction and m en participate in m arket production, This structure does not encourage fam ilies to take their daughters in school for long as a daughter at home i m ight release.her m other for incom e generating opportunities. G lick and Sahn (1997) analyze gender differences in earnings in Guinea. Their results indicate that education plays an important role in allocating labour force participants am ong sectors and that there is heterogeneity in the urban m arket and wages differences by sector. W om en are found to be less likely than men to be w age em ployees. The results tend to support findings by M eng and M iller (1995), G roshem (1990), S chu ltz and M w abu (1998). In a related stu d y fo r Rom ania, 9

16 Paternostro and Sahn (1999) find increasing returns to education and experience to be significant for both m ales and fem ales. T hey also found higher incidents of discrim ination in rural areas and at lower levels o f education (however, O razem and V odopivec (1995, 1999) em ploy a related approach to show that though wom en in Estonia and Slovenia w ere less m obile across jobs, they gained relative to men from changes in structure o f w ages and em ploym ent brqught about by transition to. m arket econom ies) These results support earlier studies, w hich argue that education is the most im portant determ inant o f differentials in earnings and labour m arket participation {Bigsten and Horton (1997), Appleton et al (1990), Berhman and W olfe (1984), Collier (1990), Knight and Sabot (1990), M wabu and Evenson (1997)}. Appleton et al (1990) argue that the gender differential in access to jobs in Cote d Ivoire is confined to. the private sector, w hich is attributed to low educational levels m apping onto lower wages and therefore on to a lower supply response. The author observes that discrim ination in tha labour m arket gives rise to three of the observed gender biases: First, controlling for education, w om en are less likely to w ork for wages than men. Second, parents are less likely to invest in the education of girls than in that o f boys. Third, wom en are less educated and hence less likely to be in th«labour markat, Using a similar approach, Maglad (1998) applies the Mincerian hum an capital earnings function to estim ate age earnings and fem ale labour supply functions fo r S udan. B igsten and H orton (1997) use evidence from Ethiopia, 10

17 Uganda and C ote d Ivoire, to show that there is a low level o f fem ale schooling due to discrim ination and biases in the education system. t I G raft-johnson (1978) exam ined factors affecting labour force participation rates in Ghana using data obtained from 1966 and 1970 census. Using data from the 1970 census, a m ultiple regression model w as estim ated. The results supported the observation that m any w om en who m igrate do so in accom panying their husbands. : T h eir m igration is not likely to be directly m otivated by econ om ic considerations. T he child/w om an ratio, a proxy for fertility, had a positive sign. For both sexes ' persons w ho have had som e form al schooling w ere found m ore likely to be in the labour force. Participation rates w ere higher in the agricultural sector. Peak activity rates occur at quite advanced ages o f 60 and 54 (70.0) according to 1970 census t and 55 to 59 (70.5) according to the 1960 census. This could be attributed to the fact that by this age women generally have completed their child care responsibilities. The proportion of.w idow s and divorcees in fem ale population also rises in later years. Standing (1978) carried out a study in Jam aica w ith the m ajor objective of identifying the behavioural determ inants o f fem ale particip a tio n 'in urban Jam aica. A stratified sam ple survey o f 540 w om en w as carried out in Kingston m etro area. The propensity to participate in, the labour force was considered as a function of respective opportunity cost o f activity and inactivity. ll

18 The independent variables affecting labour supply w ere age, fertility, hum an capital and m igrant status. All respondents w orking for one day or m ore in the past week and all those w ithout w ork w ho m ade an effort to find w ork w ere considered econom ically active. A il those not w orking w ho expressed w illingness to secure jobs w ere econom ically active. An index w as designed to indicate participation over a m ore extended period. The num ber o f hours o f w ork per w eek w as also m easured. i v '» The results noted that a young jo b s e e k e r for a given occupation and degree of skill w ould face grea ter com petition for em ploym ent than som eone older. C i M arried wom en m ay have m ore incom e security than single w om en but are also likely to have greater need for income. Young children reduced fem ale labour force participation, but migrant women had a relatively high probability of participation. Health had a pronounced im pact on labour supply while dem and for income had a consistently negative coefficient as hypothesized. Standing (1978) in another study in Sri Lanka conducted a survey on the determ inant# of fem ale labour fares participation. A linear function w as tested. The function w as based on household decision m aking behaviour in w hich the probability o f labour force participation w as determ ined by opportunity cost o f activity and inactivity. The m ain proxy variable for incom e w as level o f education since education determ ines the level o f incom e opportunity. - i J 12

19 . He used ordinary least squares. The sam ple consisted of w om en aged and living in e ith e r rural o r urban areas. It w as noted that in Sri Lanka, fem ale t participation is related to m arriage, fertility and education. A decline in the level of fertility m ay be expected to increase extent o f fem ale labour force activity. Rising levels of education could be expected to lead to greater fem ale participation but this does not mean that if all w om en had the sam e education as m en all w om en would have a sim ilarly high probability o f participation. Lachaud et al (1994) indicate that for m arried w om en the decision to participate in the labour m arket and not have leisure or do dom estic w ork is positively related to age or the level o f schooling. However, that decision is negatively related to the income, o f other m em bers o f the household if leisure is a norm al good, due to positive income elasticity. Lachaud et al conducted a comparative study based on a sample o f 6 sub Saharan African countries: Burkina Faso, Cam eroon, C ote d Ivoire,. t ' 1 Guinea, Madagascar and Mali. They indicate that the mode of employment of the head o f the household is an, im portant determ inant o f the social relationship that shapes the environm ent of labour, in particular the process of access to em ploym ent (p. 203). They conclude that in A frica fem ale labour supply is much lower than that of men, the positive correlation o f the labour supply with the developm ent is particularly explained by the differentiated participation of women in the labour m arket. I3

20 Petch (1978) conducted a study on fem ale labour force participation in Latin A m erican countries of Chile, Costa Rica, Ecuador and Venezuela. The results noted that fem ale labour force participation decreases as the num ber o f children rises. A husband s incom e affects the need for additional incom e. However, in lowincom e fam ilies, fem ale participation will be higher if they have m any children since a large num ber o f children reduces per capita resource available and m ay force the w ife to work. T he study also noted that those w ith higher incom e have higher levels ' o f education and vice versa. \, i Pang (1978) exam ined the determ inants o f labour supply in Singapore during the period 1957 to 1974 and noted the follow ing findings: 1) Education is po sitively related to fem ale labour force participation, 2) The larger the household income, the less likely will be the w om an s participation in the labour market, 3) The number of children in the household will have a negative effect on the Participation o f fem ales but this relationship w ould be different w ith the l. availability o f dom estic help. It w as noted that between 1957 and 1974 m ost im portant changes in labour supply in Singapore resulted from the rise in the fem ale participation rate.' The change can be attributed to the unprecedented econom ic grow th in Singapore during the analysed period.,14

21 W orld B ank (1980) offered several reasons fo r the disparity in em ploym ent opportunities in the m odern se cto r betw een m en and w om en. T h e m ajor reason cited vyas that the urban w om an s low econom ic status in labour force is prim arily a result o f cum ulative discrim ination over tim e in both provision o f and dem and fo r educational services. T h e study noted that education significan tly increased the probability of a w om an, obtaining em ploym ent in the form al sector. E ducation by itself did not ho w ever gu arantee em ploym ent. N ew m an (1984) expla in ed that sex differentials in school enrolm ent are m ost pronounced at age 15 years and above, the ages o f higher secondary, vocational and se co nd ary education. S ub Saharan A frica 'must ad dress itself to the latter enrolm ent differentials if w om en are to be able to acquire the skills needed for productive employment in the modern»» economy. - Neo-classical theories emphasize that women participate less than men in labour» ' force because they have low er levels of hum an capital m ainly education, training of all types and on the job experience* A nker and Hein (1986). Apker and Hein (1986) note that women can improve their participation in labour force w hen their educational level is increased. They base the hum an capital approach on the follow ing two assum ptions:, 1) W om en s labour,force participation is o f necessity interm ittent because o f their natural child-rearing role! v ' I5

22 2) Men and w om en have equal access to job opportunities and com pete on an equal basis in the labour market. Kibua (1981) noted that wom en w ho have 1-4 years-form al education exhibit highest fertility. High fertility was negatively related to participation o f fem ale in the labour force. He em phasized that the pattern o f tim e allocation at a particular time reflects the op po rtu nity cost o f tim e in alternative uses. T he op po rtu nity cost o f a : w om an w ho is not w o fkin g is grea ter on the farm and hom e than in the labour > ' m arket and an increase in her non-w age incom e w ould m ake her allocate m ore of her time to that activity. A w orking w om an s w age rate is equal to her opportunity cost o f tim e at hom e and on the farm. H er pattern o f tim e allocation depends on its effects -on the m arginal productivity o f tim e in these alternative activities. For instance, if she gains more in terms o f value of time on the farm than in wage rate Iq.the market, m ost o f her tim e will be devoted to farm production. The study also em phasized the fact that educational attainm ent Increases the productivity on the farm thus depressing the dem and fo r children. Education therefore raises the opportunity cost o f tim e so 'th a t less time is allocated to hom e and therefore child production. Kinyanjui (1.981) noted that the crucial forces in the transform ation o f the position of wom en in the society is their increased education and the changes occurring in the com m unity. He divided Kenya into different regions using the criterion of sducational developm ent. He claim ed that educational de ve lopm e nt is closely 16

23 related to the econom ic and political developm ent achieved by each region. He found a strong positive correlation between educationally advanced regions and the advance o f w om e n s education. Note how ever that his results w ere too aggregated t and could be concealing a lot o f inform ation. i 2.2. Overview of Literature Previous studies em phasized such factors as education, fertility, m arital status, household income, job experience, num ber o f children, m igration, history and age o f fem ales as determ inants o f the participation o f w om e n in the labour force. r ' / M arried w om en have m ore incom e security than single wom en; and w om en with children are m ore likely to have greater need fo r incom e. T h e past studies indicate that m igrant w om en had a relatively high probability o f participation. The presence * o f young children under the age o f 5 years reduced the fem ale labour participation. The studies noted a husband s incom e affects the need fo r additional incom e in that the higher the husband s income, the less likely will be the participation of the wire in the labour force. H igher educational levels lead to greater fem ale participation. The probability of fem ale participation in labour force increases w ith the availability of certain types o f work, the pay offered and the w orking conditions provided. A woman with a good education is more attracted by non-m anual options, w hich offers a high level o f remuneration more in keeping with her skills. A substantial increase in participation rates o f m arried wom en w ould be possible only if there are 'improvem ents in cultural changes in so ciety and a reduction in fam ily size. 17

24 G enerally speaking, the problem at hand is that o f allocation o f tim e to various econom ic activities. A nker and Know les (1978) analysed m icro level variables and m acro level variables,,» like health, unem ploym ent and average m odern se cto r wage. T he m acro urban labour m arket co nditions had a significant effect on urban fem a le la b o u r force participation. It should be noted that these studies are m ostly interested in the analysis o f the supply o f labour o f the household as a 'w h o le or an analysis o f the characteristics of supply o f m arried w om en s labour. Although these studies enable us to define to som e extent the object o f o u r work, w orks that are clo ser to our interest (w om en as a w ho le) are lacking. This research is a co ntribution tow ards that direction. It will, also contribute to the literature by exploring the K enyan case, which is under researched.' i 18

25 C H A P TE R M E TH O D O LO G Y ' ', - i 3.1 Sampling procedure In this study, G reater Nairobi has been defined according to the geographical and social limits, using the nearest neighbour criteria rather than by the adm inistrative limit criteria. The procedure used w as m ulti-stage proportional to population size sam pling. The first stage consisted o f stratification o f the adm inistrative divisio ns and extra t areas drawn from the G reater Nairobi. This w as necessary so as to get a representative sam ple o f the diverse population o f G reater N airobi in term s of socio-econom ic status and density. A d m in istra tive area of N airobi has eight divisions, fifty-one locations and one hundred and ten'sub-locations. T h e additional _» * r«a«drawn from the environs of Nairobi were treated as one division because they constitute about fifteen percent of the total population of Greater Nairobi hence the total num ber o f adm inistra tive divisions w ere nine. T h e boundaries o f G reater l Nairobi w ere adjusted using the satellite im age. I f This study Is part of a large study. The sub-sam ple o f this study w as obtained from the large study. The procedure follow ed by the large study w as as follows: In order to avoid high clustering effect it w as necessary to sam ple one hundred and fifty enum eration areas (EAS) scattered am ong the nine adm inistrative divisions. This num ber o f clu sters w as arrived at based on the experiences from countries 19

26 were sim ilar surveys have been conducted. G iven the uneven distribution of households and EAS in the Nairobi adm inistrative areas, the selection o f EAS from each division w as proportionate to the num ber o f households in each division. The follow ing table sum m arizes, the sam pling procedure follow ed fo r Nairobi City. Table 2: Sampling procedure D iv is io n N u m b e r o f EAS N u m b e r o f h o u s e h o ld s P e rcenta g e h o u s e h o ld s p e r d iv is io n Central M akadara Kasarani Em bakasi Pum w ani , 11 W estlands Dagoretti Kibera T o ta l Source: Census 1999 S e le c te d n u m b e r o f EAS The rem aining tw enty enum eration areas w ere selected from the environs outside the adm inistrative boundaries of.nairobi city using a satellite im age to draw the limits of greater Nairobi. The second stage w as to random ly select the Enumeration Area in each division from the 1999 census list of Enum eration Areas using random number, generator from the SPSS program version 9. In the third stage in each Enum eration Area, a sam ple of fifty households w as random ly selected using a system atic sam ple from the household listing from the central bureau o f statistics (CBS). This household listing w as updated prior to the survey. The fourth stage was to sam ple the biographies (individuals) drawn from the sam pled households. Generally, in African countries w here such surveys have been conducted, as m any households as po ssib le w ere draw n in orde r to get the ne ce ssary nu m be r o f 20

27 individuals in the older generations (45-54 years). This is because the age structure usually form s a pyram id in most African cities. However, the age structure of N airobi according previous censuses, (1979.and 1989; results from the 1999 w ere not available) show s greater disparities. The pyram id is highly skew ed im plying. few er fem ales com pared to m ales in each generation. In particular, there are unusually few e r w om en in the age range The distribution as per gender and! generation is as in the table below: Table 3: Distribution o ft he population by Gender and Generation Generation In percent of the total population Males Females ' Assuming that there are on average 3,28 persons per household, the expected / - - sample population in the households aimed at: > 150 (EAS) x 50 households x 3.28 persons =24600 persons. This form ed the entire sam ple but due to problem s o f non response, individuals w s re Interview ed, From those interview ed,,11693 w ere fem ales and these are the ones used fo r analysis In this study. i Model Specification and estimation technique. The study analyses the determ inants o f fem ale labour force participation using logit model. Since labour participation is a dichotom ous variable, it can be estim ated Using either a linear or non-linear estim ation m ethod. O rdinary least squares (OLS) is the m ost com m on linear estim ation m ethod. T he m ajor problem w ith it is that the estim ated co nditional probabilities m ay. not lie w ithin the logical lim its o f 0 and 1. 21

28 P rob it o r logit m odel guarantees that the estim ated conditional probabilities lie betw een 0 and 1..» In a dicho to m ou s discrete m odel, logit gives the sam e results as probit model i (M addala 1983). Logit m odel is, used because in addition to describing a discrete function such as the one w e have investigated in this study, its functional form is also sim pler. M oreover, logit is less restrictive com pared to the probit model. 4 Formally, the conditional observable probability (Pi) that; a woman is in the labour force can be expressed as, R = W here Z = XIJ + e J + e -. Where, I Pi s P robability that a w om an is in the labour force. /. / Z "= a logit index X = A vector o f the socio-econom ic characteristics that the wom an has. IS = A vector of the param eters to be estim ated 6 = A vector of error terms that follow a logistic distribution. i t The variables included in X are age o f the wom an, the square of the age of the wom an, w orking status of the spouse, num ber of children a woman has, previous em ploym ent experience, household s non-labour income, education - dum m y variables, religion dum m y, dum m y variable for m igration status and l 22

29 ' l m arital status dum m y variables which includes m arried, single and widow ed w om en dum m y variables Theoretical framework t Definition, measurement and justification of the variables used. T h is-sectio n discusses the variables used to explain the decision by wom en to. participate in the labour force, their m easurem ent as w ell as reasons for their expe cted relationships. Marital Status A w om an's m arital history is an im portant determ inant o f labour force participation. M arried w om en have m any household responsibilities and are therefore expected to participate less in the labour force. Am ong wom en, single w om en are expected to show the strongest attachm ent to the labour force. This is because these wom en do pot have m any responsibilities and thus nave abundant time to work. W om an.who are divorced, w idow ed or separated are m ore likely to show a higher degree of participation com pared to m arried wom en. This is because they are m ore likely to be prim arily responsible for the financial support o f a household w ith dependent children. In the case o f a m arried wom an, she m ay rely on the head of the househ old fo r financial support. Number of children, O f all possible determ inants of fem ale labour supply, the relationship between fertility and fa m ily size is probably the m ost studied. A t first glance, it seem s that 23

30 j. i > fe rtility' and fam ily size should be negatively related to fem ale labour force participation. Taking care o f children is a time intensive activity so it is usually assum ed that w om en w ith young children (less than 5 years) m ust either find som eone to care for the children or not work. Nonetheless, there has been significant criticism lodged against certain assum ptions im plicit in the argum ents concerning the negative : fertility-labour force participation relationship, especially in the case o f LD C S. First, the opportunity cost o f childcare is said to be low in LD C S. For m any households, there are other household m em bers w ho can help care fo r the children w hen the m other is away. But even in the situation w here there are no fam ily m em bers who can help care for the children, there is always the possibility o f hiring som eone to help. In LD C S, the w age paid to som eone to care for children tends to be quite low, / and it is this w ag e w hich constitutes the op po rtu nity cost o f fem ale labour force participation. T h is is w hy it is difficult to hypothesize the expected relationship between childcare burden and fem ale labour force participation. T he presence of - children less than 5 years of age would be reasonable measure of childcare burden. Labour force participation is influenced by ages of the children in the fam ily as w ell a s, th e ir num ber, so tw o different m easures w ere used to m easure this variable;- presence of children jess than 5 years and total number of children a woman has. E d u c a tio n An im portant reason as to w hy anyone w orks is the incom e received or generated, i from th is w ork. E ducational attainm ent is d ire ctly related to the la b o u r force l 24

31 participation o f wom en. G enerally, the m ore educated a w om an is, the higher the probability of participating in the labour force. The fact that better educated wom en tend to have a hig h e r lab ou r force participation rate than those w ith less education i t ~ probably reflects their ability to locate and retain desirable jobs w hich are m onetarily and psychologically rew arding as well as being relatively stable. \ Husbands working status H usbands incom e, to w hich his education and job status are closely linked, is another o f the factors affecting w om en's labour force participation. Although w ives w ith a high level o f education and w ho have high-incom e husbands m ay w ork for personal satisfaction rather than for econom ic reasons, (thereby obscuring the "Incom e effect") it is ge ne ra lly the case that the hig he r the husband's incom e, the less likely is his wife's participation in the labour force. Obviously, there is less i fin an cial pressure on her to do s o..w iv e s w hose husbands are un em ployed are i m ost likely to be in the labour force, w hile wives w ith husbands w ho are em ployed t are least likely to participate in the labour force. This is because the husband's incom e can support the financial needs of the fam ily and the w ife does not therefore have to. work. Age Age refers to the num ber of years a w om an has lived up to the tim e of interview.lt is positively related to participation o f fem ale labour force up to a certain age group then declines thereafter. In the younger age groups, w om en w ill participate less as they m a y 's till be attending school and are still dependent on their parents. In the m iddle ages, the y becom e productive and can now pa rticipa te fully in the labour 25

32 m arket. In the older age groups, they begin to drop out o f the labour force as they retire., Non labour income The present study uses fam ily property that generates incom e as a m easure for perm anent incom e. This constitutes fam ily incom e that w ould be available even. w hen the w om an w as not working. The highbr the non-labour incom e (perm anent incom e) received by the household, the low er is the probability o f the fem ale to participate in the labour force. This is because the fam ily's needs can be catered for using this incom e, hence, there will be less felt need for the w om an's incom e. Non labour, incom e is m easured according to presence or absence of property that generate incom e. Such property includes residential plots and businesses. Religion * / In Keny^, among A fricans in either rural or urban areas there does not appear to be any stigm a attached to w om en participating in the labour force. However, in m any M uslim societies and in certain Indian caste groups, w om en are culturally restricted from participating in the labour force. It is hypothesized that the Islam religion has a # t ' *. *. negative, influence on the participation of females in labour force, The probability that a M uslim fem ale w ill participate in the labour force is quite low com pared to the participation of a non-m uslim. This is because M uslim culture prohibits their wom en from w orking. i 26

33 Previous Employment History, * t P revious em ploym ent history refers to.whether a w om an has w orked in the past or * f not. W om en w ho have w orked in the past have a higher probability o f participating in the labour force com pared to those w ho have not w orked in the past (they are m ore likely to be in the labour force or in search for jobs). In m ost o f the cases, these w om en drop out o f the labour m arket due to som e inconveniences such as childcare, term ination o f contract, attend school and so on. An exception is when a w om an has retired from the labour market. Though the w om an will have experience, she is less likely to participate in the labour force but, generally an experienced person will get a job faster than one w ho is not experienced. Migration Status / \ r / * / ' M igrant,w om en refer to those w om en w ho w ere born out o f Nairobi but are now living In Nairobi. M ost of those people m oved to Nairobi at an approxim ate ago of 15 years, t h e reasons for m oving to Nairobi are m any but m ost com e to N airobi to look for em ploym ent. O ther reasons for the rural urban m igration could be to attend school, to jo in their husbands follow ing m arriage, to stay w ith relatives and so on. N on-m igrants are those w ho have stayed in N airobi ever since they w ere born. Migrant woman are expected to participate m ore in the labour force com pared to non-m igrants. This is because it is believed that the m ain reason fo r m igration from rural areas to urban centres is search for em ploym ent. 1 27

34 CHAPTER Introduction This chapter is divided into two sections; section one and section two. In section one, w e present the descriptive statistics o f the data used. In section two, we present the results of the-m odel. '4.2. Descriptive Statistics T h is se ction presents de scriptive statistics and cross-tabu latio ns fo r lab ou r force participa tio n w ith different characteristics such as age, hig he st level o f education r / attained, migration status of the woman, marital status, religion, women household heads and n o n-labo ur incom e am ong others. T he m ajor aim is.to provide basic inform ation and understanding of the sam ple data used., / / Characteristics of the sample Table 4.1: Female Participation in the Labour Force. L a bo ur P artloloatlen Froauanov Percent (%)_ Not in labour force ,fa In labour force Total ' , v As evident from Table 4.1 above, percent of all women were economically active at the time o f the survey. This implies that more than half o f the women in the sample were not economically active. J 28

35 Table 4.2: Female labour force participation by age( percent) L a bo ur participation/a ge Total(%) Not in labour force In labour force Total From T able 4.2 above, percent o f the w om en w ere betw een years of age. Those in the age category o f years w ere percent, w hile percent w ere between years of age. This im plies that the m ajority of those interview ed w ere in the older generation. The m ajority o f the w om en in the three age categories do not participate in the labour force. From this table, one can conclude that com pared to other age categories, people participate less in the labour force w hen they are young. This result could be due to the fact that m ost of them are still attending school-or some other training. T a b le 4.3: Fem ale _abour F o rce P a rtia l patlon and Noh-labour Incomefper Labour Those w ithout nonlabour incom e incom e Those w ith non-labour Total participa tio n Not in labour force In labour force Total Table 4.3 shows that m ajority o f those interview ed did not receive non-labour incom e (74.43% ). A m ajority o f the w om en not receiving non-labour incom e w ere not in the lab ou r force. The m ajority o f w om en receiving non lab ou r incom e also i participate in the labour force. From these results it is evident that m ajority of the people did not have' non-labour income, and those w ith non-labour incom e, m ajority w ere in. the labour force. The reason fo r this could be that those not in the labour 29

36 m arket m ay lack capital to invest in property that generate incom e. T h ose w orking have incom e w hich they can save to acquire property. T a b le 4.4: L a b o u r P a rtic ip a tio n and E d u c a tio n! p e rc e n t) E ducation level L a bo ur partic pation Total No education Primary level Secondary level College.level University level Total Table 4.4 show s that m ajority o f the respondents either had secondary school or college education as the highest level o f education attained (42.03 percent had secondary education w hile percent had college training as the highest level of r, / education attained). As evident from the table, at least everyone w ho was, interview ed had som e form al education. This show s that w om en in g re a te r Nairobi - ' ' i - have a high literacy level. O nly percent o f those interview ed had prim ary / - 1 duoation. Thoaa w ith university education form ed 9,31 percent o f th e labour force. / T h e -table also show s that m ajority o f the people in labour force have college education, A lso, there is a high unem ploym ent am ong se co nd ary school leavers. % «T h is m ay be because new entrants in the lab ou r m arket take tim e to adjust the ir expectations to the type of jobs available. 1 Table 4.5, Migration Status and Labour Partlcipationf percent) Labour participatiorvhiigration status N on-m igrants M igrants Total Not in labour force In labour force Total From Table 4.5, migrant women form percent of the total sample while the rest are non-migrants. The majority of the women are non migrants and are not economically active. I 30