Predicting inflation component drivers in Nigeria: a stacked ensemble approach

Area of Science:

• Economics
• Econometrics
• Machine Learning

Background:

• Inflation in Nigeria presents a complex challenge.
• Understanding the drivers of inflation is crucial for effective economic policy.

Purpose of the Study:

• To disaggregate inflation components in Nigeria.
• To identify the key drivers of headline, food, and bread & cereal inflation.
• To evaluate the effectiveness of current Consumer Price Index (CPI) weighting systems.

Main Methods:

• Utilized a stacked ensemble machine learning approach for robust inflation prediction.
• Analyzed out-of-sample test data to validate predictive accuracy.
• Disaggregated inflation into its constituent components.

Main Results:

• Food CPI is the primary driver of both urban and rural headline inflation.
• Bread and cereals are the most significant drivers of food inflation.
• Specific items like biscuits, agric rice, and garri white are key drivers of bread and cereal inflation.
• Some major inflation drivers are assigned lower weights in the CPI basket.

Conclusions:

• Focusing solely on CPI weights without identifying underlying drivers may hinder inflation control.
• Accurate inflation management requires tracing and tracking inflation sources to the sub-component level.
• Machine learning offers a powerful tool for detailed inflation analysis.