*WONKY POST ALERT* Many economists and reporters use the wrong data when talking about multifamily housing starts. Their preferred dataset shows multifamily construction starts on the rise in 7 of the past 15 months. But no one within six degrees of separation of the multifamily development business believes that to be true.
What is it, and why is it wrong?
Headlines almost always refer to what's called the "seasonally adjusted annual rate." It comes with good intentions. The goal is to annualize a single month's data so you can directly compare one month to another, but doing so requires adjusting for seasonal differences. The problem is that (at least in the case of multifamily starts) the resulting data is pretty much garbage -- wildly volatile, non-predictive and therefore useless.
That's why we continue to see crazy headlines about rising starts that don't align with what we see/hear on the ground.
Case in point: The February 2025 data released last week shows multifamily starts at 370k units, which was one of the highest numbers in the last 12 months and also comfortably higher than much of the 2017-19 period prior to COVID. It was the latest in a zig-zag trendline that, when charted, looks like sound waves. The whole purpose of seasonally adjustments is to compare one month to another, and yet the result is this since the start of 2024: Down, up, down, up, down, up, up, down, down, up, down, up, down, up.
No joke! How useful is that?
Answer: Not at all. Furthermore, it's misleading. There are no other indicators (especially from the more granular private sector data providers) to suggest multifamily starts have done anything but move consistently downward over the last couple years.
So what should we look at to properly gauge starts? Well, if you're going to use the Census, I'd use NON-seasonally adjusted starts on a trailing 12-month basis -- which also accounts for seasonality by taking in all 12 months.
In the dataset, multifamily starts in the year-ending February 2025 totaled 333,100 units, the lowest since July 2014. That passes the sniff test and aligns with the higher-quality private sector data.
Furthermore, in that dataset, multifamily starts are down in 13 of the last 14 months. And in 21 of the last 24 months. That’s a much more accurate picture of what’s actually happening. Also: If you worry a T-12 number is too lagged, you can also just compare starts in one month to the same month in the prior year. You'd get a very similar storyline.
That’s what headlines and policy wonks should be focused on.
It's important to get this right. Because if you're a policymaker or a reporter who doesn't understand that multifamily starts continue to plummet, then you're missing the real story -- and the chance to solve for it.