By by Calculated Risk on 3/15/2022 09:51:00 AM
Way back in 2013, I wrote a post “Predicting the Next Recession. This post was in response to several recession forecasts (that were incorrect).
In that 2013 post, I wrote:
3) An exogenous event such as a pandemic, significant military conflict, disruption of energy supplies for any reason, a major natural disaster (meteor strike, super volcano, etc), and a number of other low probability reasons. All of these events are possible, but they are unpredictable, and the probabilities are low that they will happen in the next few years or even decades.
Unfortunately, in 2020, one of these low probability events happened (pandemic, emphasis added), and that led to a recession in 2020.
We’ve seen several policy errors over the last several years, mostly related to immigration and trade during the previous administration, but none that would lead the economy into a recession.
And this most common cause of a recession is the current concern. With inflation picking up due to the pandemic (stimulus spending, supply constraints) and now due to the invasion of Ukraine, the Fed will embark this week on a tightening cycle to slow inflation.
One of my favorite models for business cycle forecasting uses new home sales (also housing starts and residential investment). I also look at the yield curve, but I’ve found new home sales is generally more useful. (See my post in 2019: Don’t Freak Out about the Yield Curve)
Click on graph for larger image.
The arrows point to some of the earlier peaks and troughs for these three measures.
The purpose of this graph is to show that these three indicators generally reach peaks and troughs together. Note that Residential Investment is quarterly and single-family starts and new home sales are monthly.
New home sales and single-family starts turned down last year, but this was because of the huge surge in sales and starts in the 2nd half of 2020.
The second graph shows the YoY change in New Home Sales from the Census Bureau.
Note: the New Home Sales data is smoothed using a three month centered average before calculating the YoY change. The Census Bureau data starts in 1963.
Some observations:
1) When the YoY change in New Home Sales falls about 20%, usually a recession will follow. An exception for this data series was the mid ’60s when the Vietnam buildup kept the economy out of recession. Another exception was the recent situation – we saw a YoY decline in new home sales related to the pandemic and the surge in new home sales in the second half of 2020.
2) It is also interesting to look at the ’86/’87 and the mid ’90s periods. New Home sales fell in both of these periods, although not quite 20%. As I noted in earlier posts, the mid ’80s saw a surge in defense spending and MEW that more than offset the decline in New Home sales. In the mid ’90s, nonresidential investment remained strong.
If the Fed tightening cycle will lead to a recession, we should see housing turn down first (new home sales, single family starts, residential investment). There are other indicators too – such as the yield curve and heavy truck sales – but mostly I’ll be watching housing. (I’m not currently on recession watch)
