Egyptian Married Women Don t desire to Work or Simply Can t? A Duration Analysis. Rana Hendy. March 15th, 2010

Transcription

1 Egyptian Married Women Don t desire to Work or Simply Can t? A Duration Analysis Rana Hendy Population Council March 15th, 2010

2 Introduction (1) Domestic Production: identified as the unpaid work done to maintain family members and/or a home. The classical dichotomy of work in the market vs. leisure does gross injustice to the female. Female s time allocation should better be analyzed in terms of a Three-way division: work in the market, work at home and leisure. The domestic production remains a main feature of sexes inequalities.

3 Introduction (2) Economic Interest Understand gender inequalities within the family. Gender inequalities within the family could affect: - The females participation rates in the labor market (the labor force size). - The economic growth (Stephan Klasen; 1999). - Gender inequalities outside the family: discrimination etc..

4 Paper s Contribution In developing countries: very few studies. The present research is one of the first exploring the extremely biased division of labor within Egyptian households. Why Egypt: - Participation rates of Egyptian females remain very low: New Explanations. - Females sole occupation in Egypt is housewives. - Evolution of Marriage, education, fertility and time saving devices: What are the implications of females time use. - Panel Data to observe this evolution (ELMPS ). Limitation: Data are only available for children and adult females. Males/ Females comparisons are not possible.

5 What I do This research aims at exploring females time allocation household production activities (time spent on housework, child care and market work) Analyze the main determinants of Females time allocation. Using a multi-state multi-spell duration model, this research allows a better understanding of the endogenous relation between marriage and different employment states.

6 Outline Time Use Data Time Allocation in Egypt Methodology Empirical Results Conclusion

7 Egyptian Time Use Data The Egyptian Labor Market and Panel Survey: individuals are followed on an eight-year period In the 2006 survey, a whole section is devoted to time use of Egyptian women: Domestic activities were classified into 14 groups. In the 1998 survey: Three questions on the three main activities on which the females spent the most of her time at home.

8 Sample Selection All females aged between 16 and 64 years old. In the panel analysis: we keep all females that were singles in 1998 and follow their marital 2006 in 2006 (got married or remained single). Panel sample: A total of 1850 females.

9 Descriptives Table: Time Uses of Egyptian Married Females: Sample Means by Number of Children Number of children Zero One Two or more Total Age 44,47 25,3 35,14 36 Time Uses (hours per week) Market work 7,61 6,43 10,08 8,83 Domestic work 32,74 53,91 51,15 46,72 Housework 31,9 36,23 39,65 37,04 Child care 0,84 17,68 11,49 9,68 Total work 40,35 60,34 61,23 55,5 Observations

10 Data Descriptives Figure: Time allocation by marital status

11 Figure: Evolution of Domestic work by marital status:

12 Figure: Evolution of Market work by marital status:

13 Duration Analysis First, we run Kaplan-Meier estimates of the survival function of the durations of employment and marriage separately. We assume here that marriage is exogenous. is a method of generating tables and plots of survival or hazard functions for event history data.

14 Figure: Kaplan-Meier Survival Function: Censored Data

15 Duration Analysis In a second stage, we estimate a bivariate probit relating both decisions of marriage and employment. y 1 a is equals if the female is employed and zero otherwise. And, y 2 a represents the second decision of getting married. The latter equals to 1 if the female is married and to zero otherwise (Picchio, 2008).

16 Duration Analysis Let us first consider the first spell: Let s assume that the individual occupies state κ in It is an employment spell if κ = e, or an inactivity spell if κ = u. And we denote U κ the duration of the spell occupied by the individual in We are interested in the impact of getting married during the eight-years period (1998 and 2006) on the conditional probability that this spell lasts at least D years. We note T κ the dichotomic variable describing the marital status (married versus single). Let U κ = X κβ κ + T κ δ κ + ɛ κ and T κ = Z κγ κ + ν κ be the correspondent latent variable.

17 Duration Analysis and U κ = 1, ifu κ = X κβ κ + T κ δ κ + ɛ κ D, (1) U κ = 0otherwise, (2)

18 Duration Analysis and T κ = 1, ift κ = Z κγ κ + ν κ 0, (3) T κ = 0otherwise, (4) where (ɛ κ, ν κ ) are i.i.d. N(0, Σ), X κ and Z κ are vectors of explanatory and instrumental variables respectively (κ = e, u).

19 Duration Analysis A history of a given female can be represented by a sequence of realizations of a discrete time stochastic process Y t, t 1998, 2006, taking its value in a discrete-state space E = 1, 2, 3, 4. Y t is the state occupied by the female at time t. Let us assume that the realizations of the process are independent and identically distributed. This is a discrete-time discrete-state labor market participation process (see, Fougere and Kamionka, 2008; Heckman, 1981; Lancaster, 1990). The estimation is done using the simulated likelihood estimator.

20 Duration Analysis For the specification of the model, many problems occur. First, the initial time t = 1998 does not correspond to the date of entry into the labor market for all the females in the sample. We then have to consider the problem of initial conditions in estimating the discrete time stochastic process. Two approaches can be used here: 1 The first approach is proposed by Heckman (1981) and consists on modeling the initial conditions. 2 The second approach is proposed by Wooldridge (2005) and consists on modeling the unobserved heterogeneity. In the present research, we opt for the second method because of its flexibility. Unobserved heterogeneity to be added.

21 Duration Analysis To put into nutshell, we aim to estimate: 1 The conditional probability of being married. 2 The impact of marriage on the risk to leave employment and on the employment duration. 3 The impact of marriage on the duration of non employment and the return to employment.

22 Results 1 Main determinants of Females time allocation: - Both domestic and market labor supply increase with the level of education. - The number of children increases the domestic work and decreases the market work. - Females in Cairo and Alex do more (domestic and market) work relatively to other regions. - Being part of a household that owns a family projet increases the market labor supply. - The domestic work (and specifically child care) increases with the current household welfare (only basic services).

23 Tobit Results: Time Spent on Domestic Work Singles (16-64) Married (16-64) All Coefficient Coefficient Coefficient age 2,617*** 1,412*** 4,049*** age square -0,034*** -0,023*** -0,047*** Age at marriage - 0,212** - HH whealth in ,360*** 0,891** -1,009*** Basic Services in ,641-0,753* -1,622*** Educ dummy2 14,185*** 1,349 7,925*** Educ dummy3 3,723-0,802 10,953*** Educ dummy4 8,473*** 1,865 12,243*** Educ dummy5 9,460*** 2,141 12,764*** Educ dummy6 6,654*** -1,913 9,463*** Number of Children in HH 1,454* 4,890*** 6,138*** Parent in the HH - -30,323*** -24,164*** Sister/Bro. in law ,280*** Mother in law - -15,750-39,366*** Dummy for Family Projet 1,021-1,559-0,891 Region dummy 2 0,667-7,404*** -4,042*** Region dummy 3-0,551-4,176*** -1,887*** Constant -22,570*** 22,121*** -41,202*** Sigma 19,494 27,028 25,718 Log Likelihood -6868, , ,573 Pseudo R squared 0,0188 0,0109 0,0626 N

24 Tobit Results: Time spent on Market Work Singles (16-64) Married (16-64) All Coefficient Coefficient Coefficient age -0,350 0,429 0,103 age square 0,003-0,006* -0,002 Age at marriage - 0,097 - HH whealth in ,212-1,159*** 0,368*** Basic Services in ,510-0,120-0,026 Educ dummy2-1,542-0,387-0,016 Educ dummy3 20,031-3,974 3,616 Educ dummy4 1,412-1,896-2,061 Educ dummy5-1,482-2,592-3,407*** Educ dummy6-8,518** -1,659-4,390*** Number of Children in HH 2,459-0,517-1,053*** Parent in the HH (dummy) - -14,810-0,362 Sister/Bro. in law ,277 Presence of a Mother in law ,724 Dummy for Family Projet 0,637 1,305* 1,855*** Region dummy 2-3,816** -2,988*** -3,678*** Region dummy 3-8,665*** -4,823*** -6,308*** Constant 67,175*** 61,426*** 62,445*** Sigma 13,104 11,787 12,939 Log Likelihood -1244, , ,6716 Pseudo R squared 0,0418 0,0253 0,0267 N

25 Conclusion Results of the duration analysis (in progress) would allow a better understanding of the effect of marriage on employment transitions of Egyptian females. Married women tend to not be able to reconciliate family and professional lives (Double burden). Politics already in place: Decrease the illiteracy rate.. reduce discrimination in the labor market etc.. Call for policies that would encourage women to continue working after marriage in order to increase females participation rates: - Part-time jobs - Programmes de gardes d enfants (child care institutes). - Work from home jobs.

26 THANK YOU FOR YOUR ATTENTION