Reports from Union of African Population Studies / L'Union pour l'Etude de la Population Africaine Num. 11, 1995

Union for African Population Studies, Rapport de Synthese / Summary Report, Numéro/Number 11, Mar 1995

Programme de petites subventions pour la recherche en population et developpement

Female Employment and Fertility in Selected Ethiopian Communities: A Microeconomic Analysis

Asmeron Kidane (PhD)

Professor of Applied Statistics, And Associate Member, Demographic Training and Research Centre, Addis Ababa University, Faculty of Science, P.O. Box 1176 - Addis Ababa, Ethiopia

ETUDE REALISEE DANS LA CADRE DU PROGRAMME DE PETITES SUBVENTIONS POUR LA RECHERCHE EN POPULATION ET DEVELOPPEMENT FINANCE PAR LE CRDI, LA FONDATION ROCKEFELLER, LA FONDATION MAC ARTHUR, LA SAREC ET LE MINISTERE FRANCAIS DE LA COOPERATION

Code Number: uaps95002

TABLE OF CONTENTS

PART ONE

TITLE OF CONTENTS

1. Introduction
2. Supply and Demand for Women's Labour
2.1 Determinants of Supply
2.2 Determinants of Demand
3. Women's Education and Employment
4. Fertility and women's Employment
5. Existing Evidence
6. Qualification

PART TWO

7. Survey Methods and Results
8. Indicators of Fertility
9. Descriptive Results
10. A Microeconomic Model of Fertility and Women's Employment
11. Variables Included in the Model
12. Empirical Results for All Women
12.1 Regression Results for All Women
12.2 Regression Results by employment Category
13. Conclusions and Policy Implications
14. Acknowledgement
15. References

PART ONE

INTRODUCTION

The first part of the study summarizes the conceptual and methodological issues on the relation between women;s employment and fertility. as microeconomic framework is the basis for analyzing employment-fertility relations the discussion will be confined to the review does not consider the relation between the two variables directly. The supply and demand for women's services are first summarized, this is followed by considering the relation

between female education and employment. After considering these variables we proceed to the major objective of our research that is the relation between female employment and fertility.

Most of the literature that consider employment-fertility relation and that use microeconomic framework tend to be highly technical, and most of the relation are expressed in mathematical formulation. However, in this summary, we have to present the basic issues in non-mathematical or non-technical form so as to expand the readership of the study. However, the mathematical formulation of the relation between female employment and fertility will have to be reinstated when we proceed to empirically implement the relations.

2. The Supply and Demand for Women's Labour

2.1. Determinants of Supply

There are several studies that aim at analyzing the determinants of women's labour supply; however most of them are only tangentially related to demographic factors especially to levels and trends in fertility. Within the microeconomic framework, studies by Mincer and Cain (?) focus on the supply characteristics of married women. The authors state that for married women, the alternative to employment in the market my be unpaid housework; the latter may include bearing and raising children; in other words leisure may not be the alternate to employment outside the house. A decision on whether married women will enter the labour force involve income and substitution effects of market work versus unpaid work including child bearing. When considering married women, income will refer to total family income; this is because total family income will influence the choice between employment outside and within a house.

There is generally an inverse relation between household's income and wife's potential earning may also affect the decision to work outside. When the effect of these two variables are compared Mincer states that the wife's potential income is more important than the husband's current earning. In other words there is the predominance of substitution (wage) effect than the income (husband's earning) effect. It should be noted, however, that the effect may depend on the type of data; desegregated or household or survey data tend to provide a more reliable information, even though it may not be representative. The relation between employment of married women and income may vary by residence such as rural versus urban, place or origin such as local people and migrants as well as the type of employment, such, as, formal and the informal sector. In all the cases, the level of education exerts considerable influence on women's employment, and one may find it difficult to separate educational level from labour force participation of women. Marital instability may also have an effect on women's employment and fertility; in other words women in unstable union could have high probability in being engaged in work outside the house. Wife's age may also affect supply because it may explain whether a woman has younger of older children. The effect of this on fertility will be considered later.

2.2. Determinants of Demand

So far we have dealt with the supply side; it may also be in order to consider the demand for women's services. Oppenheimer (1968) states that one of the major explanatory variables that affect the demand f or women's services is the use of labour saving devices as well as replacement for home produced goods within the household. This determinant is more pronounced in many developed countries. The fact that a major portion of GDP is from the service rather than production of goods also contributed to the increased demand for women's services. Like the determinant of supply age of women may also affect the demand for their services, the type of employment may also affect demand for women's services.

The preceding supply and demand analysis consider employment of both married and unmarried women, and the choice is between working at home and being employed outside for remuneration. Working at home has an effect on fertility. Thus, if one is to relate women's employment and fertility, one has to classify women's activity at home with those in the market. Within the household, the bearing and raising children should feature prominently in this exercise. This will be considered in detail in the next section.

3. Women's Education and Employment

Various studies show that human capital investments that are allocated by household members in less developed countries are gender specified, and favour more males than females (Gender 1989). However, such discrimination is more prevalent in societies where the income and other endowments of households are low. Such gender specific allocation of resources disappear as the income and other earnings of households increase. From microeconomic theory point of view, this may imply that the demand for daughter's human capital investment will be booth price and income elastic compared to a son's human capital investment, and such elasticities may converge as income rises or as the cost of education decreases. This argument also implies that if household's income decreases substantially, and if the cost of schooling increases, suddenly then it will be the daughters that will be negatively affected compared to a male offspring. This is likely to have a depressing effect on employment opportunities of women, and as a result, fertility levels may remain high.

This aspect of gender biased investment in human capital is part of parents life cycle economic consideration. In other words one considers the motives behind families' investment in their children's education, parents may have two consecutive periods to consider. The first part of the cycle is when parents work, and the second is when they retire. during the first period, household's consumption may be estimated as income less the a mount invested on human capital or on children's education. during the second period, consumption may be estimated as transferor remittances from children. Transfers from children to parents depend on their earning. The later may again depend on their human capital endowment or level of education. From this argument it follows that parents value their own consumption as well as their children's education. While deciding to invest on children , therefore parents have trade-off between their own current or first period consumption vis-a-vis their future consumption or children's wealth.

This type of scenario may be prevalent in developing countries where families are poor and capital markets are weak or non-existent. In the absence of capital markets investing on own children's education is one of the few options open to households. As families are poor they do not have large savings that would enable them to diversify investment but only to invest on children. This argument may also imply that return from investment in human capital may b larger compared to alternative investment opportunities.

When we consider gender differences in human capital investment the preference of investment on males over females may not be an innate characteristic of households but parents respond to market opportunities. In other words, the predominance of males in various types of employment, such as, government and the private sector would lead parents invest on sons rather than daughters. Thus, given limited family income, the preference of investment on males over females may be based on economic consideration. Because of this consideration, employment opportunities of women outside the household may be low and this is likely to keep fertility high.

It should be obvious to the reader that the reason as to why we treated women's education and employment first; it is only after considering the education-employment first; it is only after considering the discuss female employment and fertility. This is done in the next section.

4. Fertility and Women's Employment

Many developed countries which are in the final stage of the Democratic Transition are characterized by increase in the labour force participation rate of women as well as a decrease in completed fertility. The policy implication of this phenomenon for developing countries is that the labour force participation rate of women must be increased if we have to lower fertility and population growth rates. In other words, the inverse relation between women,s employment and fertility presupposes that there may be a conflict between employment outside the house and child rearing. On the other hand, McCabe and Rosenzweig (1976), state that one may have to take into consideration some unique characteristics of developing countries before the policy implication of the inverse relation between women's employment and fertility is to be considered. In other words, the peculiar characteristics of developing countries may show that employment and fertility may not be incompatible after all. It is only after identifying these distinguishing characteristics of developing countries that one can develop a framework for studying the relation between women's employment and fertility.

Some of the distinguishing characteristic of women's employment outside the house in less developed countries include

(a) Lack of substitution to mother's activity within the household.

(b) Sex specific division of labour in employment outside the house. These include the predominance of women employed in the informal sector such as retail occupation and cottage industries within the household. Employment in these and related activities could be compatible with on the job child care;

(c) In many developing countries child care could be assumed by relatives and older children, especially daughters.

In general these unique characteristics may reduce or even reverse the inverse relation between women's employment and fertility.

Because of the above characteristics, the use of general household theory have to be modified in order to study the relation between women employment and fertility. The model will be one period comparative static similar to those developed by Benporah (1973a); Willis (1973); McCabe and Rosenzweig (1976); Rosenzweig and Evenson (1976); Duraisamy (1989); and Kidane (1994). We assume that there are two types of service flows that give satisfaction to the household. McCabe and Rosenzweig refer to these as commodity services and child services. These two "commodities" are produced within the household. To produce them one needs the time inputs of husband' wife, children and goods purchased in the market. The latter is consumed by the household itself.

Adult members of household allocate part of heir time on economic activity such as formal or informal employment and this employment will enable the household to obtain market goods. The remaining time of the household will be allocated to activity within the household. The model also specifies the wages of employed husbands, wives and children. These wages will play a major factor when households decide to allocate their time between household activity and economic activities outside the house. For example, if a woman's market wage offer increases there may be two possible outcomes.

(a) An increase in the "price" of children or the opportunity cost of bearing and raising children compared to that of other commodities consumed by the household. This is because raising children requires more of a mother's time.

(b) An increase in the total income of household.

The effect of outcome "a" is referred to as a substitution effect; in other words this outcome may increase or decrease the number of children ever born depending on the amount of that a woman spends bearing and raising children. The effect of outcome "b" is referred to as an income effect; this will increase the demand for more children assuming that the latter is not an inferior good.

Taking into consideration the substitution effect, that is, the response of fertility decision to mother,s wage increase, if a woman spends more of her time in child raising compared to other activities, then such substitution effect may outweigh the income effect. If the opposite is the case then an increase in wife's wage will increase the fertility level.

(a) The ability to use purchased inputs for wives' time in rearing children; such input may include the hiring of domestic servants.

(b) The availability of older children who are willing to take care of younger children when the mother is undertaking other activities.

(c) The compatibility of some employment activities with raising children.

The above factors which may influence the fertility of behaviour of married women may differ from one society to the other. They may also differ by residence and other socioeconomic and cultural example which may affect wife's time allocation in child raising is the length of breastfeeding and the availability of marketed baby food. This phenomenon is a good example of purchased inputs that may affect a woman's time allocation between the two competing activities.

There are also others factors that affect a mother's time allocation between activities within a house and outside. It should be noted that the "production" of children as well as other activities within a house is also influenced by the ratio of the price of wife's imputed time to that of market inputs. The latter includes the services of maids. If there is a greater substitution between wife's and other persons time in child raising compared to the production of non child services, then the low imputed wages of servants and the availability of adults to care for children my induce a wife to seek employment outside, but may not reduce the levels of fertility.

Other factors that could reduce the inverse relation between female employment and fertility include the low returns from human capital whereby adults who complete certain level whereby adults who complete certain level of education have low probability of getting employment. We have already noted that in both developed and developing countries women have less chance of securing employment in the formal sector. This may induce households to keep children at home and this may be a substitute for wife's time in child rearing.

The allocation of time to child rearing compared to other commodity services could also be less whenever wife's employment is not incompatible with child rearing. The micro-economic model of fertility behaviour of households that is applied for developed countries assumes incompatibility between female employment and fertility. The model may thus have to be modified so as to incorporate the possible compatibility between the two. However, the modified model will have to recognize that this possible compatibility may also be limited to the informal and not to the formal sector. Subsequent empirical implementation of microeconomic model of household behaviour relating to fertility and women employment may have to consider formal and informal sector separately.

The number of children desired by the family, the time allotted by women for participation in employment outside the house as well as the decision to send children to school are endogenous; or choice variables and all of them could have similar determinants. For example, female wage rate has a positive effect on labour force participation rate; this in turn may affect fertility, the latter in turn depends on several socioeconomic factors and market characteristics mentioned above. In general, a simple negative correlation between women's employment and fertility that is characteristic of many developed countries. The various points discussed above should be taken into account. In the subsequent studies we will develop a model that tries to incorporate the stated issues and we will empirically test it using survey data.

When developing a model on the determinants of women's employment and fertility, one needs to find a set of exogenous variables that affects each variable separately. The preceding discussion suggests that this is not easy to do. In other words it may not be possible to find exogenous variables that influence women's employment but does not influence fertility and vice versa. In econometric estimation such impossibility would make the model under-identified and thus impossible to empirical implementation. Rosenzweig and McCabe (1976), suggests that one needs to identify estimates of wife's as well as husband's wage rate so as to jointly consider female employment and fertility. Even if one obtains female employment and fertility it may imply that w omen who wish to have more children may spend less time in employment outside the house. This may result in less experience in labour force; this again suggests lower wages and the cycle may continue. On the other hand, women who wish to spend more time in employment outside the house may have less children but more experience, and hence higher wages, and again a different cycle would continue.

5. Existing Evidence

Several empirical investigation on the relation between women's employment and fertility have been conducted; as expected the theoretical foundation of the model depends on the educational background of the investigator. A sociologist or a technical demographer or an economist may have different models and could have similar empirical results but with different models and could have similar empirical results but with different interpretation and different pox;icy prescriptions. Many of the empirical results involve simple comparison of parity levels or age specific fertility rates for the two different sets of women employed and unemployed. These studies are also controlled for the level of eduction, residence (rural versus urban) as well as the overall income of the household. Such undertaking show that employed women tend to have less children than those who are not (Gendell, 1976), except when a comparison was made between rural and urban areas. In the former, fertility seems to be positively associated with female employment (Goldstein, 1972).

A more detailed empirical investigation by Nerlove and Schultz (1970); and Schultz (1974) tried to isolate the effect of changes in women's employment on fertility. The study generally shows a negative relation between female labour force participation rate and fertility. However the strength of the negative relation depended on units of measures of fertility, (crude birth rate, age specific and total fertility rate, general fertility as well as the number of children ever born) measures of labour force participation rate of women and type of employment. The results also depended on the type of data; national, district level or survey based data, on household. The strength of the negative association between the two variables also depends on whether the data is cross-section, times series or the pooling of the two data set. In other words, it may be difficult to reach a conclusion that an increase in female employment would always decrease fertility.

A more robust result was obtained by Rosenzweig and McCabe (1976). A microeconomic model of the household behaviour which was described above is implied in the empirical investigation. The methodology proposed by the authors tries to overcome some of the econometric problems. The authors use the number of children overborne (CEB) the number of annual hours worked (AHR) as measures of fertility and female employment respectively. Before relating fertility and employment, determinants of female and male wage rates were determined. This is consistent with the discussion in the previous section where wage and the level of education will have to be related first before employment and fertility can be treated. In both equations, wage is said to depend on the level of education and age of the individual. The latter is treated as a non linear is treated as non linear determinant of the wage level. The predicted wage equations were then introduced in the second stage regression, where the predicted wage rate entered as u explanatory variable in the determinant of children overborne (CEB) as will as in the determinant of female employment (AHR). In the second regression equations besides wage rates, age of women, type of occupation, husbands education, total income of the household, religion various indicators of individual tastes as well as the use of contraceptives were included. The results indicate the importance of women's education in the relation between fertility and employment. Of particular significance is the importance of education in affecting the knowledge and use of contraceptives. The latter is expected to reduce excess number of children. The final results tend to show a negative relation between fertility and employment, but the explanatory powers of the equation was not strong.

6. Qualification

The preceding discussion on the relation between women's education, employment and fertility behaviour can only be considered within a single period comparative static framework. Such an exercise may have its own shortcomings. This is because over a period of tim, the process of family building or raising of children could vary from a situation outside the control of parents to a new situation whereby the decision to have more or less children may be guided by the household's own n preferences and desires. This is what is commonly referred to as time dependent shift from "fate" to "design". In this research we will assume that the population or area under study has reached a level whereby decision to have more children falls within the framework of design category. In other words, households are expected to make a conscious and well thought decision about the number of children they would like to have.

When deciding on the number of children, families may also simultaneously make decision on the health and education of children as well as t he labour force participation rate of all family members. In other words, if and when a decision is made to reduce fertility, households may increase their aspiration for their children health and education, and this in turn may attract women into the labour market. Inspite of such a design and intentions, it may not be always feasible for families to control the conception and timing of children's births as well as the desired sex composition; naturally they are unable to control or predict the child,s innate potentials. In other words, the attainment os the desired number, sex composition, spacing and quality of child may not be achieved to the satisfaction of parents (Lloyd 1989). Even if families succeed in getting the desire number and sex composition of children, and also can support them in the future,children themselves may not survive. This is another uncertainty that families face.

Household preference with respect to families' size and sex composition may also change over time. This may be in response to new events may be manifested within the household, or they may be a reaction to new socioeconomic and political environment. Choices that are optimal at a point in time may not be so after a lag of several years. The goals of husbands and wives may diverge over time; this will have to be renegotiated and old tastes and preferences revised. Unanticipated changes in socioeconomic conditions such as changes in the cost of education, health, overall cost of living as well as employment opportunities for men and women may change. All these unanticipated changes cannot be incorporated into the already developed model of women's employment and fertility. Such an exercise would complicate matters and make the research undertaking unmanageable. Nevertheless readers should be aware of such unpredicted outcomes.

PART TWO

7. Survey Methods and Results

The data needed for this study was collected from a Sidamo region of southern Ethiopia. The reason for choosing this area is purposive. In other words the ethnic, religious and cultural composition of the study area is more diversified, the level of education is more representative and the age distribution follows the standard pattern. 340 women were selected through random sampling and interviewed, however 312 of them had all the requisite information for a proper analysis of the various hypothesis identified in this study.

The questionnaire is meant to generate basic information necessary for the study. These include the age of household heads, and spouses, their age and age at marriage, place of residence, ethnic and religious affiliations. It also included age and age distribution of children overborne, type of employment of husbands and spouses, years of experience and earnings. Other sources of income of wealth as well as subjective questions on the actual and desired age at marriage, actual and desired length of breastfeeding as well as the decision making process of households. Because of budgetary constraints, only questions that have direct relevance to proposed study have been included in the questionnaire. A descriptive statistics indicate that the value of the variables are within acceptable limit, and that one can proceed to a detailed analysis of the various hypotheses and the proposed microeconomic model.

Table 1: Descriptive statistics of Selected Variables For

8. Indicators of Fertility Potential of the Area

In this study a measure of fertility is the number of children overborne (CEB). The rationale for the choice of this fertility variable over the others is because it is direct and easy to measure. Other fertility measures such as crude birth rate or general fertility rate do not seem to be appropriate for survey data. A brief description of the number of children ever born classified by the number of women is given in table 2. The results indicate that the mean number of children ever born is 2.9; this is irrespective of the age mother. The results suggest that this is indeed a l ow value. However, a clear picture of fertility behaviour of the study area can be shown when the number of children ever born is cross-classifies with the age of the mother. This is given in Table 3. The result in this table shows that only 20.8% of t he 15 to 19 years women have no children which is

Table 2: The distribution of women by CEB

Table 3: The Distribution of Women by age-groups and CEB

an indicator of high fertility potential in the area. This is reconfirmed by the fact that as much as 22.4% of 20-24 years old mothers have between three and six children. More statistics on the high fertility potential of the area can be extracted from Table 3.

9. DESCRIPTIVE RESULTS

We have presented seven possible hypotheses so as to relate woman's employment and fertility. Each hypotheses will be tested by presenting simple and cross-classified tables. In the next section we will develop an appropriate microeconomic model of employment and fertility relations, and try to verify whether the seven hypotheses have some validity. we will then compare our findings from descriptive statistics and those from the model so as to check the internal consistency as well as the robustness of our finding.

In order to test the seven hypothesis the sampled women are classified into three categories, namely:

(a) Those who are not employed

(b) Those employed in the informal sector

(d) Those employed in the formal sector

The above classification will be necessary for an objective assessment of the stated hypotheses. Out of the total number of 312 respondents 169 women (54%) are not employed, while 86 (27.5)

and 57 (18.5%) of the respondents are employed in the informal and formal sector respectively. This allocation is not predetermined. After making a selection of 312 households on the basis of the procedure described earlier it was found that the classification of respondents into the three employment categories earlier. The seven hypothesis will be tested for each of the three employment categories.

Hypothesis 1

There is a trade-off between women's employment and fertility; however the extent of the trade-off may depend on the type of employment.

In order to verify this hypothesis we cross-classified number of children overborne with age of mother for the three employment categories. This is given in Table 4.1 to 4.4.

Table 4.1: The Distribution of Women with No Employment By Age- Group and CEB

Mean CEB=2.63S.D=2.05

Table 4.2:The Distribution of Women Employed in the Informal Sector by Age-Group and CEB

Mean CEB=3.13S.D=2.28

Table 4.3: The Distribution of Women Employed in the Formal Sector

by Age-Group and CEB

Mean CEB=2.75S.D=2.02

Table 4.4: Mean and Standard deviation of CEB among Younger and Older Mothers

The mean number of children overborne among employed women seems to be slightly less than those formally employed. A comparison between those employed in the informal sector and the other hand, the only six per cent of unemployed women have no children. The corresponding values for women employed in the informal and formal sector is seven and eleven per cent respectively. In other words, there is higher tendency to control fertility among women in the formal sector. A clear picture emerges when one compares mean CEB among young (twenty five years or less) mothers with older mothers. (more than 25 years). The results clearly shows that fertility is less among older women employed in the formal sector compared to the other two categories. There is no significance difference between unemployed women and those in the informal sector. Thus hypothesis 1 seems to have some validity; that there is a trade-off between fertility and employment in the formal (not formal) sector.

Hypotheses 2

Given that employment is classified into formal and informal employment, there is higher probability of trade-off in the formal employment compared to the informal employment. Table 4.4. shows that this hypothesis is sustained.

Hypothesis 3

Within the informal sector (such as trade), there is some trade-off between employment and child bearing at an earlier age of woman's reproductive period. Such a trade-off may become weaker at later ages; that is during the period when there are older sons and daughters who may look after newly born brothers and sisters.

Hypothesis 4

Within the formal sector, the trade-off is more likely to be stronger at latter ages of a woman's reproductive life' The reason in this case is different from that under the informal sector. In the formal sector household's desired number of children may be met relatively early; this would then enable women to be engage in formal employment.

Hypothesis 3 and 4 may be verified by referring to Table 4.3 and Table 4.4... For example within informal employment, 15% of women have more than five children after the age of 30. The corresponding value for women employed in the formal sector is only eight per cent. Thus a strong trade-off between employment and fertility within the formal sector at later age seems to prevail; this is not the case within the informal sector. The result also shows that here is age CEB relation among informally and formally employed women is lower than those employed in the informal sector.

An alternative approach of verifying hypothesis 3 and 4 is to cross classify CEB with age at first marriage. The rationale here is that women who are employed in the formal sector may be required to reach a certain level of education. This in turn may delay the age at first marriage; such a requirement in informal employment. The age at first marriage fertility relation is shown in 5.1 to 5.3

Table 5: Age at First Marriage and Fertility by Type of Employment

Table 5.1: The distribution of Unemployed Women by Age at First

Marriage and CEB

Table 5.2: The Distribution of Women Employed in the Informal Sector, by age at First Marriage and CEB

Table 5.3: The Distribution of Women employed in the Formal Sector, by Age at First Marriage and CEB

As expected the mean age at marriage among unemployed and informally employed women is almost similar; that is 17.93 and 17.56 respectively. On the other hand, the corresponding value for women employed in the formal sector is 19.86. also 65% of unemployed women were married when they were 18 years or less. Similar estimate (64%) is observed among women employed women in the informal sector. For formally employed women the percentage is only 44. This is an indirect indicator of strong trade-off between fertility and type of employment.

Hypothesis 5

The trade-off between women's employment and fertility is much stronger if the wage rate or other remuneration is high.

In an attempt to verify this we have classified fertility with income for both the informally and formally employed women. This is given in Table 6.1 and 6.2. It was already noted that fertility of women in the informal sector was lower than those in the formal sector. At the same time earning of women in the informal sector is lower than those in the formal sector; while the average earning of women employed in the formal sector is about 400 birr those in the informal is only about 100 birr. There is also much variability in the earnings of women in the informal sector suggesting the unpredictable nature of their earnings. More than half of women in the informal employment earn less than 100 birr while more than 300 birr. Thus hypothesis 5 seems to have some validity.

Table 6.1: The Distribution of Women in the Informal Sector by Spouses' Earnings and CEB

Mean=71.60SD=174.54

Table 6.2: The Distribution of Women in the Formal sector by Spouses' Earnings and CEB

Mean=306.62SD=272.41

Hypothesis 6

The trade-off is weaker if the husband's earning is high and if the household is relatively wealthy. Information on the wealth of household was either unreliable or unreliable. Thus we have to relate husband's earning with fertility for the three categories of women. This is shown in Table 7.1 to 7.3.

Table 7: Husbands Earnings and Fertility by Type of Employment

Table 7.1: The Distribution of the Unemployed women by Husband's Earnings' and CEB

Table 7.2:The Distribution of women Employed in the Informal sector by husbands' Earnings and CEB

Mean Earning=185.39SD=173.22

Note that this table is the same as table 6.1

Table 7.3: The Distribution of Women Employed in the Formal sector by Husbands' earnings and CEB

Mean Earning 485.94SD=370.28

One may observe that there is a significant difference in husbands' earnings among the three categories of women. Husbands' of women employed in the formal sector earn more than twice compared to the other two categories of women. Also husbands' of women who are not employed earn more compared to the husbands of informally employed women. The correlation coefficient between husbands' income and fertility is -0.17, 0.12 and 0.09 for unemployed, informally and formally employed women respectively. This result suggests a stronger trade off between husband's earning and fertility for women employed in the informal and formal sector. It should be noted that negative correlation suggests a stronger trade-off. Thus hypothesis 6 tends to have some validity.

Hypothesis 7

Educated women are more likely to make a trade off between employment and child rearing than less educated women. This hypothesis was treated by cross classifying fertility and education data. This is given in table 8.1 to 8.3

Table 8: Women's Education and Fertility by the Type of Employment

Table 8.1: Distribution of the Unemployed Women by Level of Education and CEB

Mean Educ. = 8.22S.D=3.04Correlation=0.33

Table 8.2: The Distribution of Women Employed in the Informal Sector by Level of Education and CEB

Table 8.3: The Distribution of Women Employed in the Formal

Sector by Level of Education and CEB

We have already noted that there is high trade-off between fertility and women employed in the formal sector compared to the other category of women. The table above shows that women in the formal sector are more educated with a mean of 10.64 years compared to the other categories of women. Even though all categories of women show negative correlation between fertility and educational level, the impact of the latter on the former is more pronounced among the group of women who are employed in the formal sector.

10. A Microeconomic model of fertility and Women's Employment

The theoretical model that we will be using in this study is based on the microeconomic model of the general household production framework. When one considers a microeconomic model of household behaviour related to fertility and employment, one assumes a single period comparative static framework, and considers family demographic and economic behaviours. These microeconomic models have three components, which are described below:

(a) A Utility Function where the variables that affect individuals and household's level of satisfaction are identified in broad terms while emphasizing fertility and employment behaviour of households.

(b) A production Function where dependent variables are the variables on the right hand of the utility function, while the explanatory variables are time and goods purchased from the market.

(c) An income and time constraint

The objective is to maximise the utility function subjects to production, income and time constraints and expressing the final function in reduced form.

In this exercises the household utility function can be expressed as

U = U (N, Y,µ ) .......................(1)

where

U = a utility function that is expected to possess the desirable properties. In other words the functions is assumed to be continuous, twice differentiable and concave

N = Number of children

Y = a composite variable that measures other consumption goods

µ = a taste parameter

N and Y are assumed to be produced within the household. A linear homogenous production function is assumed. To produce N and Y, one needs time and goods purchased in the market. In other words the household production function of good j is given by

j = f:(X_j, T_iJ;E)............................(2)

Where j = N,Y

X_j = market purchased goods needed for the production of

commodity j

T_ij = time required by household member to produce commodity j . Household member may be father (f), mother (m) and

child (c). In other words j = m,f,c

E = is an environment variable

The production of commodity j is constrained by time and household income. Thus the time constraint can be expressed as:

Ti = T_im + T_ic + ΣT_ij ..........................(3)

where Tim = total time available

Tim = tim spent in market activities

Tic = time spent in leisure

Tij = time devoted to the production of j

Family income constraint is given by

V + ΣWiTi = ΣpjXj .............................. (4)

Where i = m,f,c and N, Y

V = non labour income

ΣWiTi = total income formed by father, mother and children

(Wage = Wage rate)

ΣPjXj = total expenditure spent on commodity Xj(Pj = price ofXj)

One should note that the income expenditure constraint in equation 4 assumes that all income is spent.

If one maximises the fertility function in (1) subject to the constraint expressed in (2), (3) and (4) one gets the following reduced form equation.

j=Ψ (p_i, W_m, W_f, V, E, α) .................... (5)

Since our objective is to relate fertility and women's employment equation (5) is estimated for j = N only. In other words;

N = N (P_i, W_m, W_f, W_c, V_f, E, α) ............... (6)

One can now use equation 6 to make prior predictions. The following possible results may emerge

(a) Increase in Non Labour income (v). This may have pure income

In other words, an increase in V will have a positive effect on N provided that children are not "inferior commodities". At the same time a high non labour income will increase the

opportunity cost of woman's time and could reduce the probability of woman's employment outside the house.

(b) Wage of mothers W_m Before one is to consider the effect of mother's wage on fertility and employment, one will have to assume that raising children is mother's time intensive. Also one should assume the compensated price effects should dominate the income effect. Under this assumptions a high

mother's earning will have negative effect on fertility while at the same time inducing mothers to work outside the house.

(c) Wage of father W_t An increase in father's wage, could have positive effect on fertility and this would in turn reduce the probability of women's employment.

(d) Wage of children W_c A higher wage of children will increase fertility and this would imply low probability of women,s employment outside the house.

(e) Taste E There are different proxies for taste; these include; age, migration status, place of birth, religion and ethnic affiliation.

(f) Environmental variable α A good measure of environmental variable is the availability of older children relatives or servants within the household.

11. Variables included in the Model

There is usually wider divergence between the variables that are present in the theoretical model and those that are presented for empirical implementations. The set of explanatory variables used in the study as well as their definitions are given below:

Wage Variables

EMPW = Employment status of women

1 = employed

0 = not employed

WAGEM = monthly earning of husband

WAGEW = monthly earnings of wife

NONLAINC = remittances non-wage income, number of cattle owned or a combination of the three.

Taste Variables

AGEW = age of wife (mother)

EMPW = Whether wife is employed

1 = if employed

0 = otherwise

PLACEB = Place of Birth

1 = if non migrant

0 = if migrant

RELIG = RELIGION

1 = if orthodox

0 = otherwise

ETHNIC = 1 = if Amhara

0 = otherwise

Environment variable

RELSERV = Number of older children, servants or relatives in the house.

The variable age is entered into the model by assuming that the relation between fertility and age is non linear. In other words fertility increases with age but at a decreasing rate as the age of mother reaches 30 years or more.

12. Empirical Results

12.1. Regression Results for All Women

Table 9 shows an Ordinary Least Square (OLS) estimate of the determinants of fertility. Two of the significant explanatory variables that affect fertility in the expected direction are husband's or spouses' earnings. As expected wife's earning has a negative effect on fertility. This result is consistent with the prior prediction of the model as the opportunity cost of mother's time at home increases with increase in her earnings. In other words the price effect far outweighs the income effect, and this in turn induces mothers to allocate more of their time working outside and less on childbearing and rearing. An increase in the husband's wage would have positive effect on fertility. This positive income effect is likely to reduce the probability of women's employment. Even though the sign of non-labour income seems to have expected positive sign and that this variable increases fertility the coefficient is not statistically significant. Also the relation between women's employment status and fertility showed the expected negative sign even though it is not statistically significant.

Out of the four 'taste' variables, education seems to have a significant negative effect on fertility. In other words, women with a higher level of education tend to prefer less number of children. This is likely to reinforce the expected positive relation between education and employment outside the house. In the model, we had assumed that fertility and age of mothers have a non linear component. While the linear component of age is positive and significant, the non-linear component is negative but not significant. The expected positive sign between age and fertility may also have in indirect positive effect on woman's earning. This is because age is a good measure of on the job experience, and older women may earn more and this is likely to reduce fertility. This is consistent with the negative sign of non linear component. Other taste variable include: place of birth, ethnicity and religion. The prior sign of the coefficients cannot be determined in advance; two of these three variables have significant coefficients. Migrants seem to have less preference for more children than people born in the study area. This is intuitively appealing because migrants are usually better educated than local residents and the effect of education on fertility have already been discussed. Followers of the Orthodox church have a higher probability of bearing and raising more children than followers of the other religions. Also women of the Amhara ethnic group in the study area are likely to have more children than other ethnic groups. Most Amhara women are also followers of the Orthodox Church.

The "environmental variables" namely the availability of relatives or servants within the household is likely to have a positive effect on fertility; it was already noted that, women,s employment outside the house and higher fertility amy not be inconsistent for women employed in the informal sector. The estimated equation shows the positive and significant relation between fertility and the availability of servants and relatives with the household, even though women of the three employment categories are included in the model In general, eight out of the twelve explanatory variables - including the constant term - are significant, the adjusted is quite high for cross section with prior expectation. The results in the Table 9 could be considered as the major findings of the study.

12.2 Regression Results by Employment Category

The same model was applied to each employment category of women. This approach certainly reduces the degree of freedom of the model especially among women employed in the formal sector.

Table 10 shows OLS estimates for 169 women who are not employed. In this model, 5 out of the eleven variables are significant. Other, have the expected signs even though the coefficient are not significant. For example non wage income and husbands salary and the presence of relatives or servants tend to have positive effect on fertility. The effect of education and age are the same as those presented in Table 9. The same is true with the three variables - place of birth, religion and ethnicity, even though the latter in not significant. The explanatory power of the equation is also high and highly significant.

Table 11 shows OLS estimates for 86 women who are employed in the informal sector. Most of the coefficients are similar to those in Table 9, the major difference being that the "environmental" variably seems to be significant at 20%; this suggests that employment and fertility are not incompatible in the informal sector. Table 12 shows OLS estimate of 57 women who are employed in the formal sector. Because of the lower sample size and low variability of CEB in the age structure, the results do not seem to be satisfactory. Only the environment variable and religion seem to be significant variables.

The relation between fertility and female employment was also considered by controlling the age of the mothers but excluding age from the set of explanatory variables. To this effect a two separate OLS estimates were made for women less than or equal to 25 years old; and for those above this age group. The results are shown in Tables 13 and 14. For younger mothers education and religion were found as being the most important explanatory.

In other words, more educated and non-Orthodox Church mother tend to have lower fertility. The availability of servants or relatives within the household, and the status of women's employment showed the expected sign, even though these variables were non significant.

Some interesting results emerged when we estimated the model for older women. In other words, the prior signs and the level of significance of the important income, and taste variables give the expected sign. When comparing the two lasts OLS estimates, the following observations can be made:

(a) Among employed women there is a delayed age of child bearing and a conscious attempt at reducing fertility.

(b) The husband's and spouse's wage play an important factor at later ages.

(c) Education plays a major factor for both age groups

(d) Child bearing is reduced dramatically at later ages

(e) Ethnicity and religion seem to play a major role at later ages.

13. Conclusions and Policy Implications

The preceding discussion shows that the perceived inverse relation between women's employment and fertility may not be applicable to any type of employment category. One may conclude that women employed in the informal sector do not show substantial decrease in fertility. The empirical results from the Ethiopian setting suggests that the employment-fertility relation among women employed in the informal sector is similar to those who are not employed at all. In other words, while women employed in the formal sector show not seem to be the case among women employed in the informal sector. In other words, there is the incidence of compatibility between fertility and women's employment in the informal sector. This compatibility comes about because there amy be older children especially daughters, relatives and servants within the household who may take care of newly born children while mother is working outside. At the same time it is not uncommon for women to work in the informal sector, such, trade with a child on their side or their back.

A second major variable that affect women's employment is the opportunity cost of women's time outside the house and in the formal sector, while the corresponding high earning of husbands may induce women to stay at home and thereby increase the fertility rate. At the same time, employment in the formal sector is dependent; in other words, not only does higher education induce women to wok outside the house, it also delays age at marriage and introduces women to the technologies of birth control. It is already noted that women employed in the formal sector have relatively higher level of education, get married relatively late and abruptly reduce fertility after the age of 30. This is not the case with unemployment women and those employment in the informal sector.

The policy implication of the above findings suggest that the inverse relation between employment and fertility may be realized if women are relatively more educated and they have marketable qualifications. Existing government sector such as handicraft and small scale cottage industry, while engaging women in productive and income generating activities are unlikely to have a depressing effect on fertility. In such a scenario, the income effect is likely to dominate, provided that children are classified as "normal goods"

Even though employment and fertility are likely to move in the opposite direction as the stage of socioeconomic development reaches a higher level, it is education that is the most important variable that affects fertility not only by enabling women to engaged in activities outside the house but also by delaying age at marriage; by introducing women to various aspects of fertility control; by enabling households to make a conscious and well sought decision regarding the number of children they desire to have; and by making proper allocation of the available time between activities within and outside the house. Generating employment opportunities of women in the informal sector while the latter's level of education is still may not yield the desired decrease in fertility.

14. Acknowledgement

This study is financed by Union of African Population Sciences. The author would like to than Dr. Ibrahima Diop and Professor A. B. C. Ayayo for their encouragement. The results and the issues raised in the paper do not reflect the views of the funding organisation.

15. REFERENCES

• BEN PORAH, Y. (1973a) Economic Analysis of Fertility in Israel: Point and Counterpoint Journal of Political Economy 81(2) 5202-233.

• Duraisamy, P. (1989) Fertility and Child Schooling in Rural India

• Evidence From Aggregate Panel Data. A paper presented at a Conference on the Family Gender Differences and Development. Economic Growth Centre, Yale University.

• Kidane, A. (1994) A model of Economic-Demographic Interaction within a Household. A paper present at the International Symposium on Economic Modelling. The World Bank, Washington D.C.

Copyright 1995 - Union for African Population Studies.