Female Employment, Earnings Inequality and Household Well-being: The case of urban Turkey

Note: Robust standard errors in brackets. ‘*’ denotes significance at 10%, ‘**’ at 5%, and ‘***’ at 1%. First stage results relating to the participation equation are not shown for brevity.


Table A2: The contribution of subcomponents to inequality including imputed wages for unpaid family workers

Factor	SCV	Share in total income (%)	Proportionate contribution of factor (%)	Per unit contribution
All sample
Male earnings	2.30	54.3	45.53	0.84
Female earnings	18.79	9.43	13.78	1.46
Other earnings	816.59	1.35	9.41	6.97
Non-wage income	3.63	34.93	31.28	0.90
Bottom quintile
Male earnings	0.46	63.22	61.35	0.97
Female earnings	9.96	4.91	2.41	0.49
Other earnings	119.51	0.49	0.55	1.12
Non-wage income	1.49	31.37	35.70	1.14
2^nd quintile
Male earnings	0.46	56.33	31.24	0.55
Female earnings	10.27	4.51	12.14	2.69
Other earnings	97.81	0.53	0.93	1.75
Non-wage income	0.96	38.62	55.70	1.44
3^rd quintile
Male earnings	0.44	55.81	46.77	0.84
Female earnings	7.84	5.20	18.62	3.58
Other earnings	121.26	0.43	2.6	6.05
Non-wage income	0.86	38.56	32.0	0.83
4^th quintile
Male earnings	0.43	55.64	54.25	0.98
Female earnings	5.59	7.50	18.80	2.51
Other earnings	173.7	0.32	0.34	1.06
Non-wage income	0.86	36.54	26.62	0.73
Top quintile
Male earnings	1.38	51.93	42.33	0.82
Female earnings	6.89	12.87	13.31	1.03
Other earnings	226.0	2.28	13.66	5.99
Non-wage income	2.44	32.92	30.70	0.93

1 SIS web-site: http://www.tuik.gov.tr.

2 Hours of work refer to the main job held. Normal and actual hours worked are not distinguished. If the individual holds an additional job, the hours worked in that job are added to his/her weekly total.

3 In the rest of the paper we refer to such women as ‘working’ women though by no means do we imply that women engaged in reproductive work are not working. This is simply done to ease discussion.

4 Among the most commonly cited papers on the topic are Rao (1969), Fei, et al. (1978), Shorrocks (1982), Lerman and Yitshaki (1985), and Silber (1989). See Cancian and Reed (1998) for an extensive review.

5 The CV is defined as the ratio of the standard deviation to the mean of a variable. One-half the squared CV is equivalent to the Generalized Entropy measure with a parameter value of 2.

6 Furthermore, Jenkins (1999) has turned the calculation of the proportionate contributions of factors into a straightforward exercise by introducing the program to be executed on STATA.

7 The rest of the distribution is as follows: 19.1% in the second quintile, 14.9% in the third quintile and 15.5% in the fourth quintile.

8 The wage equation estimates used to predict unpaid family worker’s wages are reported in the Appendix. There are a total of 520 unpaid female family workers. In 40 cases, the assigned wages has caused their contribution to surpass total household earnings, which indicates that their earnings must have been over-estimated. The above figures exclude 20 such households.

9 It should be mentioned that in the vast majority of the households (97.5%) there is only one working woman. In 74.9% of the remaining cases - where there are multiple female participants - women share the same employment status.

10 There is further evidence of positive assortative mating (in earnings) in our data. In the sub-sample of households with non-zero female and male earnings, the correlation coefficient between the two quantities is 0.44.

11 In this section, income shares are computed by taking simple averages of sample figures. In previous sections, the shares were computed for each household and averaged through, i.e. they were weighted averages. Therefore, the shares reported here can be slightly different from the ones reported earlier.

12 In the analysis above, female earnings only include non-imputed real earnings of working women. When the same analysis is repeated by imputing wages to unpaid family workers, results do not change and therefore, are not shown in the text (see Appendix Table A2).