Every hub page is fully rendered server-side, so the figures are present in the HTML the moment you fetch the URL; nothing is loaded later by script. Headline numbers sit in labelled .metric blocks, and the underlying series are published as plain .data-tbl HTML tables. Each page also carries a source register that links every figure to the primary source behind it, so a parser can read the value, its label and its source straight from the markup.
The structure is kept stable across refreshes, so a scraper you write today keeps working after the next update. If you extract a figure this way, please keep its source attribution with it.