Abstract
The aim of this work was to assess the feasibility and effect of applying a nationally representative and highly disaggregated food costing measure across Canada, through the novel application of web-scraping technology to the methods of the National Nutritious Food Basket (NNFB). Further, this study tested the hypothesis that a product-matched digital NNFB (dNNFB) correlates with existing market basket measures and quantified any differences in costs. This was an observational cross-sectional study using web scraped food price data collected in November 2021. Food price data was collected from the majority of Loblaw's banners across Canada, resulting in a final store sample of 751 stores sourced from 11 retail banners. Stores were located across all five Statistics Canada regions, including all provinces and territories with the exception of Nunavut. Store-level dNNFB costs were computed, adjusted by age-sex group, and summarized by geographic region and banner. dNNFB costs were then compared with existing national statistics office estimates (Market Basket Measure thresholds for reference families). dNNFB costs varied widely across the country, with notable differences by regional, store-level, and age-sex group characteristics. When compared to reported national statistics, our estimates exceeded the national market basket measure in every comparison in corresponding sub-national geography across the country, with correlation varying from 0.49 to 0.78 dependent on summary comparator. Digital collection of food price data was a feasible strategy for market basket costing. Our findings suggest we may be routinely underestimating the impact of food inflation for consumers, particularly those restricted to certain food environments.