Housing Market Indicators

The lock-in effect, and the year the model broke

Standard housing analysis assumes higher rates reduce both demand and supply of transactions, cooling prices. After 2022 the demand half worked and the price half did not, in several markets at once.

The reason was the existing stock of fixed-rate mortgages. Millions of households held loans fixed at 3% or below. Moving house meant giving that up for a 6.5% loan on a similar property, which is a large increase in monthly cost for the same amount of house. So they did not move. Listings collapsed, and with supply falling roughly as fast as demand, prices held up despite the sharpest affordability shock in decades.

The lesson generalises beyond housing. A transmission channel that has worked for forty years can be blocked by a structural feature that did not exist before, in this case a very large stock of long fixed-rate debt taken out at historically low rates. When a model stops working, the first question is what changed in the plumbing rather than whether the data is wrong.

Where housing data misleads

• Monthly volatility is enormous. Housing starts carry a margin of error wide enough that the reported change is frequently not statistically distinguishable from zero.
• Weather dominates winter prints. A cold January suppresses starts and the rebound in March is not a recovery in demand.
• Price indices lag badly. The most-cited repeat-sale indices are published around two months after the period they cover and use a three-month average, so they describe conditions up to five months old.
• Local supply is not national. Planning restrictions, land availability and demographics vary enough that a national average can hide opposite conditions in two cities.

Rent, and why shelter reaches the inflation print a year late

Housing also enters inflation, and it enters slowly. Shelter is about a third of the US consumer price basket, the largest single component, and most of it is owners' equivalent rent, an estimate of what an owner would pay to rent their own home. It is measured from a sample where each unit is revisited roughly every six months, so it reflects the whole stock of leases rather than the ones signed this month.

The consequence is a lag of nine to twelve months between market rents turning and the index following. A trader watching new-lease rent measures had a view on core inflation well before the official series showed it, in both directions. It is the clearest case on the calendar of a number that is accurate, widely reported, and describing last year.

What to watch if you trade rate-sensitive markets

You do not need the whole release calendar. Two series carry most of the information: weekly mortgage applications for purchase, which is the earliest read on demand, and building permits, which is the earliest read on supply and feeds construction employment a few months later.

Both matter beyond housing itself. Construction is a large employer, so a sustained fall in permits shows up in payrolls two to three quarters later. And because housing is the fastest channel through which rate changes reach the real economy, a stabilising housing market is often the first evidence that a tightening cycle has done its work, which is precisely the evidence that shifts rate expectations and the currencies that price off them.