ABSTRACT This study examines the changing consumption patterns of households in Pakistan from 2005-06 to 2018-19, by shedding light on the impact of changing prices and income on the consumption of various food groups. For empirical analysis, the study utilized both parametric and non-parametric approaches for examining the impact of various economic factors especially food prices on the consumption behavior of households across different regions and food groups. For parametric approach, we employed the Quadratic Almost Ideal Demand System (QUAIDS) which provides insights into the degree of substitution across various food groups whereas, for non-parametric approach we employed the Kernel Regression which allowed us to estimate the Engel’s curves for various food groups.

The findings of the study revealed significant regional differences in the consumption behavior of households. In some regions of Pakistan food commodities like vegetables, meat was found to be necessities while in other region they were considered luxuries. The expenditure share of milk was found to be more pronounced across the regions than any other food group. The regional disparities for milk are more pronounced as in the case of KPK it is a luxury food group as compared to Balochistan especially the rural area where it is treated as a necessity. Further the performance of the models was evaluated using MSE, RMSE, and R 2 which revealed that in terms of predictive accuracy the Kernel Regression model outperforms the QUAIDS. There was an exception in the case of cereals in some of the years where QUAIDS outperformed the Kernel Regression model. We suggest that the choice between the parametric and non-parametric should align with the underlying objectives of the study.