Forecasting a recession with construction data

A recent thread I did on Twitter that may be of interest to subscribers.

1/ Everyone seems to think the recent jobs numbers show a strong US economy. But if we ignore the headline numbers are look under the hood we see reliable recession indicators.

2/ Job growth came in strong in all sectors apart from tech. This includes construction which is a good cyclical indicator. Yet if you look at the data, construction job growth tends to lag construction investment growth - and the latter is currently deeply negative.

3/ Here is how the situation played out during the Great Recession. Investment growth went negative in Q3 2006, while construction job growth only went negative in Q3 2007. So, we should expect a lag of around a year - all else equal.

4/ Recently some strange dynamics have taken place in the construction sector. During and after the lockdowns, investment growth increased massively, but employment growth went slightly negative and then returned to normal. Investment growth has now been negative for 2 quarters.

5/ Because we know that there is a lag operating, we can use investment growth to forecast employment growth. Here is the fitted model together with the actual data. We see that the model breaks down in the strange lockdown period. More on that later.

6/ Based on present investment growth, construction employment should go negative in Q3 2023. But the model might be too generous here given (i) the runaway investment growth in the lockdown that had no employment effects and (ii) the weakness of the present economy.

7/ Construction employment growth is a reliable recession predictor. When we see negative double digit growth in construction we know that the economy is probably in recession.

8/ So, it is worth watching construction employment growth in the coming months. When it goes negative we will know that a recession is approaching and when it registers double digit negative growth rates we know that we are in the depths of a recession.

9/ But worth emphasising that, with investment in the construction sector deeply negative, it is likely that this is already baked in. Unless we see that investment growth rebound - which would require house price declines to reverse - the recession is almost certainly coming.

10/ Also worth thinking about the meaning of the high construction investment growth that did not produce employment growth during the lockdown. Does this lead to us overestimating the real economic impact of investment in recent years? Qute possibly.

11/ To sum up: a recession is coming. Based on a simple model we would expect it toward the end of the year. But taking into account recent anomalies and economic weakness, it may well emerge in the first half of the year.