AI Analysis: Wholesale trade, sales

Executive Summary

Canada's wholesale trade sales grew 624.7% over 33 years, rising from $14.7 billion in January 1993 to a peak of $132.3 billion in May 2023, before moderating through early 2026. The dataset covers 397 monthly observations across 46 NAICS industry categories, revealing a highly right-skewed distribution where a small number of large aggregate sectors dominate average sales. Key disruptions include a sharp -21.6% COVID-19 shock in April 2020, a dramatic +44.0% rebound in January 2021, and an anomalous mid-2022 surge that produced 9 statistical outliers above the ±2.5σ threshold.

Key Findings

• Wholesale trade sales grew 624.7% over 33 years, from $14.7 billion in January 1993 to a peak of $132.3 billion in May 2023, reflecting sustained long-run expansion in Canadian wholesale activity.
• The distribution is heavily right-skewed, with a mean of $7.6 billion nearly 3x the median of $2.6 billion, driven by large aggregate NAICS categories pulling the average upward.
• 9 statistical outliers (Z-score > 2.5) were identified, all concentrated in May–September 2022, when sales surged well above historical norms likely due to post-pandemic demand and elevated commodity prices.
• The COVID-19 pandemic caused the single largest monthly decline on record at -21.6% in April 2020, followed by the largest monthly spike of +44.0% in January 2021 as economic activity rebounded.
• Most NAICS wholesale trade categories display strong positive correlations with one another, suggesting that broad macroeconomic cycles — rather than sector-specific factors — primarily drive sales movements across the industry.
• Sales values span an extreme range — from a minimum of ~$20.7 million to a maximum of ~$138.5 billion — a ratio of over 6,600x, reflecting the mix of narrow sub-sectors and large aggregate trade categories in the dataset.
• Seasonally adjusted and unadjusted series are nearly identical in median values, though seasonally adjusted figures carry a slightly higher mean ($7.68B vs. $7.48B), with both adjustment types represented by roughly equal sample sizes of ~11,000 records each.