Went back through three years of construction insolvency data this month, from the 2,213 firms in the year before this whole run started, through the record 3,490 in the most recent financial year. The trend line is one of the more useful things I track, precisely because it's a leading indicator rather than a lagging one.
Every builder that fails mid-project doesn't just represent a lost business. It represents homes that were approved, sometimes partially built, that won't hit the market on the original timeline, often not for another year or two once a new builder picks up the site.
So when I read an approvals number that looks encouraging, I mentally discount it by however bad the insolvency trend has been over the preceding eighteen months. Right now that discount is still meaningful. It won't stay that large forever, but it hasn't fully unwound yet either.