Residential and wealth-related disparities of high fertility preferences in Ethiopia: A decomposition analysis

BACKGROUND: Fertility preference significantly influences contraceptive uptake and impacts population growth, especially in low and middle-income countries. In the previous pieces of literature, variations in fertility preference across residence and wealth categories and its contributors were not assessed in Ethiopia. Therefore, we decomposed high fertility preferences among reproductive-aged women by residence and wealth status in Ethiopia. METHODS: We extracted individual women's record (IR) data from the publicly available 2016 Ethiopian Demographic Health Survey (EDHS) dataset. A total of 13799 women were included in the study. Multivariate decomposition analysis was conducted to identify the factors that contributed to the differences in the percentage of fertility preferences between rural and urban dwellers. Furthermore, we used an Erreygers normalized concentration index and curve to identify the concentration of high fertility preferences across wealth categories. The concentration index was further decomposed to identify the contributing factors for the wealth-related disparities in high fertility preference. Finally, the elasticity of wealth-related disparity for a change in the socioeconomic variable was estimated. RESULTS: The weighted percentage of women with high fertility preference among rural and urban residents was 42.7% and 19%, respectively, reflecting a 23.7 percentage point difference. The variations in fertility preference due to the differences in respondents' characteristics accounted for 40.9%. Being unmarried (8.4%), secondary (14.1%) and higher education (21.9%), having more than four children (18.4%), having media exposure (6.9%), middle (0.4%), richer (0.2%) and richest (0.1%) wealth were the positive and city administration (-30.2%), primary education (-1.3%) were the negative contributing factors for the variations in high fertility preferences due to population composition. Likewise, about 59% of the variations in fertility preference were due to variations in coefficients. City administration (22.4%), primary (7.8%) and secondary (7.4%) education, poorer wealth (0.86%) were the positive and having media exposure (-6.32%) and being unmarried (-5.89%), having more than four children (-2.1%) were the negative factors contributing to the difference in high fertility preferences due to the change in coefficients across residents. On the other hand, there was a pro-poor distribution for high fertility preferences across wealth categories with Erreygers normalized concentration index of ECI = -0.14, SE = 0.012. Having media exposure (17.5%), primary (7.3%), secondary (5.4%), higher (2.4%) education, being unmarried (8%), having more than four children (7.4%), rural residence (3%) and emerging (2.2%) were the positive and city administration (-0.55) was the negative significant contributor to the pro-poor disparity in high fertility preference. CONCLUSION: The variations in high fertility preferences between rural and urban women were mainly attributed to changes in women's behavior. In addition, substantial variations in fertility preference across women's residences were explained by the change in women's population composition. In addition, a pro-poor distribution of high fertility preference was observed among respondents. As such, the pro-poor high fertility preference was elastic for a percent change in socioeconomic variables. The pro-poor high fertility preference was elastic (changeable) for a percent change in each socioeconomic variables. Therefore, women's empowerment through education and access to media will be important in limiting women's desire for more children in Ethiopia. Therefore, policymakers should focus on improving the contributing factors for the residential and wealth-related disparities in high fertility preferences.