Why under five children are stunted in Pakistan? Multilevel analysis of Punjab Multiple Indicator Cluster Survey (MICS-2014)
Abstract Background : Pakistan is facing an acute problem of child under-nutrition as about 38 percent of children in Pakistan are stunted. Punjab, the largest province by population and GDP share having stunting prevalence of about 33.5 percent moderately and 13.3 percent severely stunted children of less than five years. Thus, this study aims at examining empirically the determinants of stunting (moderate and severe) at different level of hierarchy. Methodology : Data for this study is coming from Punjab Multiple Indicators Cluster Survey (MICS-2014). MICS uses two-stage, stratified cluster sampling approach. MICS is sub national level (Punjab province) data covering urban and rural areas. The data consists of 25,067 children under five, for 9 administrative divisions and 36 districts of Punjab province of Pakistan. Descriptive statistics and multilevel hierarchical models were estimated. Multilevel data analyses has advantage because it provides robust standard error estimates and helps in finding variation in the data at various levels. Results : Punjab has stunting prevalence of about 27 percent moderately and 10 percent severely stunted children of less than five years. The results depict that increasing age of child, increasing birth order, illiterate mothers and fathers, lack of sanitation facilities and being poor are associated significantly with the likelihood of moderate and sever stunting. Surprisingly, there is a gender bias in stunting in Punjab, Pakistan and being girl child is more likely associated with moderate and severe stunting which depicts the patriarchal nature of the society and a strong prevalence of gender bias in household resource allocations. Conclusion : This outcome of our analysis points towards targeting not only households (focus on girls) but also their families and communities. Keywords: Undernutrition; Stunting; Child Health; Pakistan; Punjab; Multilevel Models; Multiple Indicators Cluster Survey.