General influences on women’s employment in 28 countries
70 To calculate these predicted probabilities, an equation based on the coefficients from Model 4 (Table 6.2) is
formulated. Using the averages (as reported in Table 4.2) for each of the variables and the different values for education, the logged odds are calculated which are transformed to the probabilities presented here (see e.g. Pollock, 2009).
3.10 To the extent a woman has more human capital, the greater the probability a woman is gainfully nonagriculturally
✓ employed.
3.11 To the extent the household belongs to a social network at a higher socio-economic class, the greater the probability a ✓ woman in that household is gainfully non-agriculturally employed.
Box 6.4
3.12 To the extent that the presence of women in the public sphere in a community increases, the greater the probability that ✓ a woman in that community is gainfully non-agriculturally employed.
3.13 To the extent that the dominance of traditional care roles in a community increases, the smaller the probability that a ?✓ woman in that community is gainfully non-agriculturally employed.
Box 6.5 � �� �� �� �� �� �� at least some tertiary secondary completed primary completed no completed education
figure 6.3 Predicted probability on employment (%) for an average woman in an average context
Women's employment (%)
variable (1)
Model 2, table 6.2
(2) (1) + education
Economic development: Wealth level -0.665 ** -0.753 ***
Male labour supply: Male non-employment -2.715 *** -2.465 ***
Labour market structure: Share white collar 2.161 *** 1.927 ***
Labour market structure: Share skilled labour 0.910 * 1.256 ***
Degree of urbanisation 0.674 *** 0.629 ***
Norms: Women in public sphere 3.801 *** 2.975 ***
Norms: Traditional care roles -2.410 *** -0.251
Education (ref = less than primary completed)
primary completed, secondary not - 0.585 ***
secondary completed, no tertiary - 1.481 ***
at least some tertiary - 3.169 ***
Note: These coefficients are based on a model of which all specifications are the same as Model 2 in Table 6.2. The only difference is that the dummies measuring women’s educational attainment at the micro level are included.
instead of employees, as there were no useable data available on government employee numbers, which might have shown a clearer relationship.
Matching demand and supply also drew attention to the resources of women and their households. If the resources were greater, it was expected that women would more easily find a suitable job (Box 6.4).
Education as the operationalisation of human capital is very strongly related to women’s employment. With significant logged odds coefficients of 0.4, 1.1, and 2.6 for primary, secondary, and tertiary education (compared to no education), this is one of the main factors influencing women’s employment. The two highest levels of education are especially important. As Figure 6.3 shows, the probability of employment rises most steeply between having completed secondary education and having at least some tertiary education: from 21% to 53% – at least, for the average woman living in the average district of the average country.70 However,
it should not be forgotten that few women (<7%) have enjoyed higher education. The second resource focussed on is the socio-economic status of the social network a woman has access to, in terms of her spouse’s job. It seems that women whose spouse has a lower white-collar job, a blue-collar job, or whose spouse is unemployed, all have roughly the same employment opportunities. The main differences are found for women with a partner in agriculture (even after control for living in a city) and with a husband who has an upper white- collar job. The former have a considerably lower likelihood of employment, the latter a somewhat higher one. These results are roughly in line with expectations. However, the effect is only modest: the increase in odds on employment is about 13% to 19% for women with an upper white-collar spouse, compared to the women with a partner with a lower (non-agricultural) occupational status.
Both education and partner’s occupation as indicators of household resources have been shown to be influencing women’s employment positively, but it is clear that education is the more important of the two. Hypotheses 3.10 and 3.11 are thus supported. Primary education has a positive effect as well, while others have argued that it leads to a lower likelihood of employment (e.g. Aromolaran, 2004; Kuepie, Nordman & Roubaud, 2009). Given the large number of uneducated women and the efforts to increase the basic education levels in many of the 28 countries, this is an important conclusion. Providing women with at least primary education is thus potentially likely to increase women’s employment.
In this section, the results have shown that not only is the presence of jobs important, but there should be a match between women and the jobs that are available for these women to become employed. On the demand side the presence of service sector jobs seems very important, and on the supply side women’s education.
6.6 tHe added vaLue of worK: SoCIetaL and InternaLISed norMS
Values were expected to form the background of other variables and to have direct effects. Regarding the latter, more traditional norms at the societal level (see Box 6.5) and more traditional values held by household member or internalised by women (Box 6.6) were theoretically linked to lower employment levels, and the same goes for institutionalised conservative values, more particularly, state institutionalisation of conservative Islam (Box 6.7).
At the district level, both variables measuring societal norms are related to women’s employment in the bivariate analyses, and the direction is as expected. In Table 6.2’s Model 2 the relationships are still statistically significant, but somewhat more modest due to the inclusion of the other societal-level variables. Even when not taking into account the possible indirect effect through education, the positive direct effect of the public sphere being less dominated by men is still substantial (Table 6.4 ). Women living in a district with a score one standard deviation (0.16) above its mean (0.21) have odds on employment that are 159% higher than women living in a district scoring one standard deviation below the mean.
The effect of traditionalism disappears almost completely after including education (Table 6.4). It might be that norms regarding traditional care roles do mainly prevent women
71 In the IGLS model, the effect even turns positive. This