These are a few things I threw together to get a better sense of how I think about housing.

First, I take housing units completed and compare that to Census’s estimate of the number of new households formed each year. Thus, we can compare household growth to housing unit growth. Though, there is no demolition in this comparison.


As always one of the major things to notice is how the drift upward in housing units added was small compared to the bust. This is especially noticeable, when you see that over the whole period the deviation from household formation was not that extreme.

However, household formation has already turned around and we would expect it to make up lost ground unless there is a permanent change in living arrangements.

We can also look at a similar phenomenon by taking cumulative additions to the stock of both households and housing units. Then taking the ratio. Here I use a demolition percentage of 0.35% per year for two reasons. First, it gives me the near term estimate of about 200K demolitions per year that is standard. Second, it eliminates most of the trend drift in the ratio.

When you’re building up a stock from flow data it takes a while for the thing to settle down. Nonetheless, because specific dates are likely to be of interest to folks I’ll give you the whole series.


Now we can focus on the last few decades


Now, obviously the census data is kind of junky and the huge correction in household numbers we got in 2001 brought down the ratio to unlikely levels.

Nonetheless, we can see that we are currently crossing the lows seen in the early 90s and to the extent this imputation is decent we are not that far from 1-to-1.

Combined with the slowdown in household formation that is likely to reverse this is at least suggestive that there is a looming outright housing shortage.