I noticed that housing units per adult has actually started to level off. This is interesting because total permits and total starts are still below past averages.

(Click on image to enlarge)

Source

So, maybe the new neutral run-rate for new units is less than 1.5 million annually. Maybe the need for new units is less acute than I have been saying.

But, there is a problem with causation here. More people means we need more homes, but also, a lack of adequate housing can lead to fewer people - both by limiting migration and by limiting family formation.

And, it is true that population growth has slowed. Before the financial crisis, it tended to run at 1-1.2%. Since the crisis, it's more like 0.7%. So, in a way, we solved the housing shortage, in part, by reducing population growth. If this is the new normal, then maybe 1.2 million units a year isn't an unsustainably low peak. But, if population growth, either through immigration or through family formation, returns to anywhere close to historical norms, then housing starts probably need to catch up a bit and then settle at something closer to 1.6 million units annually. (Ignore the big drop in housing/adult in 2000. I haven't taken the effort to try to account for the discontinuity in the data there. I suspect that mostly that discontinuity comes from an overestimate of the housing stock in the late 1990s, but it isn't central to the main trends I am discussing here.)

Certainly, an argument can be made that population growth through family formation has been naturally slowing so that we shouldn't expect population growth to continue at historical norms. On the other hand, there are many good reasons to counter that decline with more generous immigration policies.  And, while there is a long term downtrend in natural population growth, there was a sharp downshift that appears to have been related to the economic turmoil of the crisis and to the lack of housing growth since then. Even without immigration, it seems likely that natural population growth has declined more than it otherwise would have after the crisis.

Also, there is always an important signal here of rent inflation, which has persistently run high for the past 25 years and returned to high rates during the post-crisis recovery. That is not a signal we would see in a country where housing was being depressed by natural declines in population growth.