Researchers at the Bank of Ghana (BoG) have proposed an artificial intelligence-driven system that could help policymakers detect inflationary pressures earlier and strengthen the country’s response to changing consumer prices.
The proposed system, known as e-Inflation, uses big data and automated online price collection to generate near-real-time estimates of headline, food and non-food inflation, potentially providing an additional source of information for monetary policy decisions.
The researchers — Francis Mawuli Abude, Collins Kwesi Appiah and Julius B. Dasah — say the technology could complement Ghana’s conventional inflation measurement system by providing more frequent indications of price movements before official figures become available.
Their findings are contained in a working paper titled Measuring Inflation in Ghana: Evidence from Big Data and Artificial Intelligence Techniques, published in October 2026.
The paper examines whether online prices and artificial intelligence techniques can provide reliable estimates of inflation in an economy where official statistics are compiled through periodic price surveys.
The proposed framework automatically collects price information from 120 websites and analyses approximately 500,000 price observations covering the same 307-item consumer basket used by the Ghana Statistical Service (GSS).
To assess its reliability, the researchers compared e-Inflation estimates with official inflation figures between May 2021 and December 2023.
The results showed a correlation of 0.9941 between the system’s headline inflation estimates and official figures, suggesting a strong relationship between the two measures over the period studied.
The correlation was 0.9836 for food inflation and 0.9944 for non-food inflation.
The researchers also reported average deviations of 0.3684 percentage points for headline inflation, 0.6909 percentage points for food inflation and 0.1375 percentage points for non-food inflation.
They argue that these results demonstrate the potential of digital technologies to strengthen inflation monitoring, although further validation would be required before the system could be relied upon for official statistical purposes.
Faster information for monetary policy
A key advantage of the proposed framework is its potential to provide policymakers with earlier signals of changing price conditions.
According to the paper, Ghana’s conventional inflation measurement process relies on 47,877 price observations collected from 57 market centres, with official inflation figures typically released around the middle of the following month.
The e-Inflation framework, by contrast, is designed to capture online price movements more frequently, potentially allowing analysts to identify emerging inflationary trends before the publication of official statistics.
Such information could help the Bank of Ghana and other economic policymakers assess developing price pressures and improve their understanding of changes in food and non-food prices.
The researchers recommend that central banks and national statistical offices explore the adoption of big data and artificial intelligence techniques to supplement existing approaches to measuring inflation.
They also propose extending the methodology to other macroeconomic indicators, including gross domestic product (GDP) and unemployment, to improve the timeliness of economic assessments.
Data limitations remain a concern
Despite the promising findings, the researchers acknowledge that reliance on online prices presents challenges that could affect the accuracy of the estimates.
Some websites may display outdated prices, while others may restrict automated data collection, potentially limiting the reliability and coverage of the information gathered.
The researchers therefore stress the importance of validating digitally collected prices against prices observed in physical markets.
Such validation would be necessary to determine how well online price movements reflect actual consumer experiences across different locations and market segments.
They maintain that the proposed approach should complement, rather than automatically replace, established inflation measurement methods.
The authors also clarify that the working paper represents research in progress and does not necessarily reflect the views of the Bank of Ghana, its Board or Management.
