- Study conducted on Model-based District-level Estimates using HCES 2022–23.
- Focuses on Uttar Pradesh, using statistical modelling to estimate MPCE at district level.
- Committee chaired by Dr. Mausumi Bose to explore model-based estimation.
- Auxiliary data used included schemes like Ayushman Bharat and Antyodaya.
- Top rural districts in MPCE: Bagpat, Saharanpur, Gautam Buddha Nagar, Meerut, Ghaziabad.
- Top urban districts in MPCE: Gautam Buddha Nagar, Gonda, Ghaziabad, Bagpat, Lucknow.
- Models used: Fay–Herriot (FH) and Spatial Fay–Herriot (SFH).
Developing Local Insights Through Model-based Estimation
The National Statistics Office (NSO), under the Ministry of Statistics and Programme Implementation (MoSPI), has completed a study that paves the way for more inclusive and data-led district development in Uttar Pradesh. By using statistical modelling approaches based on the Household Consumption Expenditure Survey (HCES) 2022–23, the study has produced reliable estimates of Monthly Per Capita Consumption Expenditure (MPCE) at the district level.
Background and Framework
In response to recommendations by the National Statistical Commission, a pilot study was initiated to assess the practicality of model-based estimation methods for smaller administrative areas. A committee headed by Dr. Mausumi Bose, former professor at the Indian Statistical Institute (ISI), Kolkata, was formed to lead this initiative. The NSO and the Directorate of Economics and Statistics (DES), Uttar Pradesh, worked together to provide technical support.
The Need for Local Estimates
While the HCES offers comprehensive consumption data at national and state levels, there has been an increasing demand for statistics at the district level, especially for tailoring development programmes and monitoring welfare schemes. However, limited survey sizes in individual districts pose challenges in generating accurate direct estimates.
To address this issue, a model-based approach was adopted for Uttar Pradesh as a pilot effort, using auxiliary data to augment direct survey results and improve statistical precision at the district level.
How the Model-based Approach Works
The pilot study used Small Area Estimation (SAE) methods to combine survey findings with district-level data from administrative sources. This approach enhances result accuracy for smaller areas. Among the auxiliary datasets used were those on old age pension beneficiaries, Ayushman Bharat enrolments, domestic LPG connections, and Antyodaya food scheme beneficiaries.
Two key modelling techniques were used in this study: the Fay–Herriot (FH) model and the Spatial Fay–Herriot (SFH) model. These methods helped create robust MPCE estimates where direct sampling data was insufficient.
Key Insights from the Study
The data revealed distinct consumption patterns across Uttar Pradesh’s districts:
- Rural Areas: The highest MPCE was found in Bagpat, followed by Saharanpur, Gautam Buddha Nagar, Meerut, and Ghaziabad.
- Urban Areas: Gautam Buddha Nagar ranked first, followed by Gonda, Ghaziabad, Bagpat, and Lucknow.
The study demonstrated that model-based estimations can serve as a cost-effective way to generate reliable and scalable district-level statistics, especially in resource-constrained conditions.
Conclusion
This pilot project reveals the potential of statistical models in bridging data gaps at local levels. The findings open up avenues for improved planning, assessment of living conditions, and focused policymaking, especially for welfare programmes and poverty alleviation. The same approach can be applied to other states and indicators, such as employment or poverty statistics, empowering policymakers to implement evidence-based strategies for equitable growth.