AI Analysis: Non-bank mortgages extended, by lender type, insurance status and mortgage characteristics

Executive Summary

Canada's non-bank mortgage market expanded significantly between 2020 and 2025, with total residential mortgages extended growing 60.4% in value (from ~$47.1B to ~$75.5B) and 86.6% in volume (from 142,783 to 266,401 mortgages). The dataset — covering four lender types across 24 semi-annual reporting periods — is heavily right-skewed, with a mean value roughly 78x higher than the median, reflecting a small number of very large transactions dominating the aggregate figures. Outlier analysis identified 91 anomalous data points and one negative value, likely tied to pandemic-era disruptions and subsequent interest rate cycles.

Key Findings

• Total residential mortgages extended by all non-bank lenders grew 60.4% in value, rising from approximately $47.1 billion to $75.5 billion between early 2020 and late 2025.
• Mortgage volume surged 86.6% over the same period, increasing from 142,783 to 266,401 mortgage extensions, indicating broader market participation beyond just rising property values.
• The dataset is strongly right-skewed, with a mean value of ~$2.98 million compared to a median of just $38,306 — a 78x gap driven by a small number of very large mortgage transactions.
• Four lender types are tracked — All non-bank lenders, Credit Unions, Mortgage Investment Entities, and Other non-bank lenders — across 24 mortgage characteristics including insured vs. uninsured, originations, refinancing, and renewals.
• 91 statistical outliers were detected using the 3x IQR method across 4,608 records, with one anomalous negative value of -$95K found in the 'Other renewals, refinancing and other mortgages, uninsured residential mortgages' category, likely reflecting a data correction.
• The interquartile range spans from $2,325 to $2.17 million, meaning the middle 50% of values already cover nearly three orders of magnitude, underscoring the extreme diversity in transaction sizes across lender types.
• The dataset covers 192 unique statistical series over 5+ years of quarterly data, making it well-suited for tracking how macroeconomic events such as the COVID-19 pandemic and the Bank of Canada's rate hike cycle shaped non-bank lending behaviour.