Housing Starts, Permits and Home Sales Data

Building a house isn't a single event — it's a sequence, and the US statistical agencies publish a number for nearly every stage of it. Getting the order right is most of what you need to read this data correctly: a builder applies for a permit, then breaks ground on a start, then eventually finishes and either sells the home (a new home sale) or an existing owner lists theirs (an existing home sale). Each stage lags the one before it by weeks to months, which means each series answers a slightly different question about the housing market.

Permits are the leading edge — they're cheap to file and reversible, so they respond fastest to changes in expectations, especially interest-rate moves, since most home purchases are financed. Starts follow, usually within one to three months, and represent an actual commitment of capital and labor, so they're stickier and less prone to being cancelled on a whim. Sales come last and measure realized demand rather than builder intentions, filtered through whatever mortgage rates buyers face at closing.

Picture that curve as the ordering of the housing pipeline: adjust the parameter and notice how a change at the front (permits) takes time to fully propagate to the back (sales) — it's the same lagged-response idea, just with construction and buyer timelines standing in for the plotted variable.

Worked example

Suppose mortgage rates jump from 6% to 7.5% over two months. Permits — the fastest to react — drop sharply the same month, as builders pull back on filing for projects they no longer expect to sell easily. Starts fall more gradually over the following one to two months, since builders already mid-project don't abandon a half-built house just because financing conditions worsened; they finish what they started but slow new groundbreaking. New home sales weaken with a further lag, as it takes time for the higher-rate reality to filter through to buyers actually signing contracts, and existing home sales — dependent on current owners being willing to list and give up their old, lower-rate mortgage — can lag even further, sometimes for a year or more, a phenomenon dubbed the "lock-in effect."

What this means in practice

Because the series lag each other in a known order, a rates desk watching for a housing slowdown looks at permits first as the earliest warning, not starts or sales. Conversely, a recovery in permits after a downturn is read as the first credible sign that housing activity is turning, well before starts or completions confirm it. Single-family versus multi-family splits also matter — multi-family starts are lumpy (one large apartment complex can swing the total) and are usually stripped out when judging the underlying trend in single-family construction.