July 2020 - Recessions and Recoveries of Jobs in Construction

July 2020


Activity in the Construction Sector can be thought to be a forecast by builders of economic growth in the near future. Like the stock market, this is sometimes spot on, but not always.

Figure 1 illustrates construction jobs as a share of total jobs for the United States since 1947 and for California since 1990. The vertical red bars are the official US recessions and the yellow bars identify the period it took for total payrolls to get back to their previous peak – the recovery.

Sharply rising shares are a symptom of builder optimism. They typically occur during the expansions which are in grey. Sharply falling shares reflect weakness in the current market but also pessimism regarding how soon the market will turn around. This typically occurs in the recession, which is too late.

The thin black line is the centered 20-year average of the US data. This represents the building of the same amount but with the wild swings eliminated. Jobs above this level and getting even farther above reveal a housing bubble, most notably in California and the US from 1998 to 2005. The years from 2001 to 2005 were a time when builders could have and should have protected themselves by reducing their holdings of buildable land, and by shorting mortgage-backed securities.

Now in June 2020, the US share is a bit over 50% which might call for some small adjustment to get it to the “normal” level of around 48-49%. If it fell much farther than that, it would likely come with an attractive opportunity to buy land again, in preparation of the full recovery to come later.

Incidentally, notice how much wider are the yellow bars at the right in Figure 1. I argued in my June Forecast document that this was caused by structural changes that left some sectors with permanent loss of jobs, manufacturing being the best example. My inspection of the data led me to the conclusion that construction is not experiencing a structural change, nor are we as a nation in an overbuilt situation.

Next we can look at what is happening in the recession/recovery so far in 2020. Figure 2 Illustrates with blue bars the sectoral job losses during the recession, from February to April 2020, and with orange bars the job gains during the first two months of the recovery, from April to June, all as a percent of February peak levels. In the center (PAYEMS) are the data for total payrolls which declined by -14.5% but recovered +4.9%. The sectors to the left of PAYEMS have typically declined greater than jobs overall and have been slow to recover. The sectors to the right have typically declined less and were quicker to recover. At the left are the usual culprits: manufacturing and construction. At the right is Leisure and Hospitality. But this time has been completely different with job losses in both durable and nondurable manufacturing substantially less than jobs overall and with the biggest problem sector being Leisure and Hospitality, followed by other Services, including for example, nail salons.

The Construction sector in the oval at the left had job losses of -14.2% that were just a bit less than the losses overall but have a +8.0% recovery in only two months which is more than half of the losses. This has had the effect of increasing the construction share of total jobs, a rather remarkable feature of a the data four months since the cycle peak.

This is not at all like 2005 when the builder risk was enormous. The year ahead will surely be troubled for services delivered face-to-face, but construction and the sale of building can be done with limited covid-19 exposure. The concentration of the covid-19 harm on our poorer citizens will probably have a serious impact on the sales of buildings designed for that group.

Figure 1. Construction Jobs Share of Total

Figure 2. Job Losses in the 2020 Feb to April Recession and Job Gains in the Recovery from April to June

UCLA Anderson Forecast