Contract talks between the UAW and GM have broken down and a nationwide strike is now at hand, the first since the 1970s. If the strike lasts a month or longer, it could take 1 percentage point off of real GDP in 4Q, or more, even if activity ramps back up in December. Lost income can also have a negative impact on consumer outlays in the months after the strike resolution since the income lost during a strike is gone permanently.
The ML auto analyst John Murphy notes that the last major strike against GM was back in the spring/summer of 1998, which lasted 54 days and was a local strike, aimed at a plant in Flint, Michigan. The 1998 strike involved 5800 UAW workers walking off the job but over the 2-month span more than 150,000 manufacturing workers were eventually impacted, as activity in the parts supply chain were shut down due to the drop-off in demand. The costs may be considerably more severe than in 1998 given that that was a local strike and this one could go national. We have not seen strikes of that magnitude since the early 1970s, when there was a work stoppage that lasted 100 days.
It is not expected that this strike will be as long, since the losses on both sides could be considerable, but the risks are there nonetheless. So below we discuss the impact of the 1970 and 1998 strike.
Strikes impact economic data
The 1970 strike, which ran from September 15 to November 20, had a considerable impact on economic activity perhaps a 1.5 to 2.0 percentage points in 4Q and some lingering impact even beyond. A recession was already in progress so the strike likely helped to intensify the impact. For this reason the analysis of the 1970 strike is particularly resonant at this time. Currently, economic activity is very sluggish so the fallout could be graver than if the economy was operating closer to full capacity, as was the case in 1998.
Payroll declines increased by about 130k on average over the period and the unemployment rate rose by 0.5 percentage points (and never returned to previous trends). Real personal disposable income declined by 0.2% per month, more than a percentage point slower than in the previous two months. Note that the BLS counts striking workers as still employed but workers laid off as a result of a strike are listed as unemployed. Manufacturing production also took a big hit. Industrial production declined by 1.1 percent on average through that period, and durable goods shipments dropped by 2.6%.
The strike also filtered into consumer outlays, retail sales were down 0.4% per month through the strike period and real PCE growth fell by 1.1% QoQ (annualized) in 4Q 1970 in comparison to about a 2.6% annualized pace in the first three quarters of the year. Retail auto sales declined by 6% MoM on average, as inventories of vehicles dried up. This took 2.6 percentage points off of 4Q GDP growth, though there was a quick rebound in 1Q. There was some slowdown in the consumption of nondurables and services expenditures as the impacted workers cut back on outlays, but that occurred in the subsequent quarters, in the first half of 1971.
1998 strike also impacted economic activity
The 1998 strike also played havoc with the data. There was an abrupt slowing in nonfarm payroll growth, with the auto sector shedding 130k workers. If layoffs were of this magnitude, we could see outright declines in nonfarm payrolls, perhaps in the October data (this is now beyond the September nonfarm reference week). The unemployment rate rose by 0.2 percentage points over the June/July period but then fell back to previous trends in the subsequent months. We did not see much of a decline in the average workweek, but there was a 1ppt slowing in industrial production nonetheless.
Durable goods orders saw an abrupt drop off, falling by 22% annualized in the 3 months up to July whereas it had been running at a better than double digit pace prior to that. Retail sales also took a significant hit; retail sales, saw outright declines of 0.5% m/m on average in July and August, due to a lack of auto supply. Even ex autos there was a pronounced slowdown in retail sales growth, as those striking workers cut back on outlays, and the net impact on real PCE was around 1.0 to 1.5 percentage points in 3Q.
The net impact on GDP took about 1 percentage point off the top-line number or $22 billion in 3Q 1998. Indeed, the strike played havoc with the numbers throughout 1998, with the auto sector adding $5.7 billion to inventories in 1Q and then running them down in 2Q and 3Q and finally rebuilding in 4Q.
A protracted strike could shut down activity quickly
This time around, the auto makers have been very careful about building up stockpiles, reducing inventories by almost $5 billion since the beginning of the year. Production through the late spring and early summer of 2007 was elevated, and we believed at the time they were doing so in anticipation of a strike. But the inventory to sales ratio for the retail motor vehicle and parts sector (where dealer auto units are captured) was at 2.13 months in July, which is slightly above the 5 and 10 year trend so stockpiles are probably only a bit elevated relative to future demand. In the wholesale sector (where a good portion of auto parts are stored), inventories are at just 1.36 months, which is low by historic norms. This suggests that any sort of protracted strike could shut down activity fairly quickly and we could see the negative impact on the key data (payrolls, claims, production, durables) as early as the October prints.
There are several other factors to note. First, the jobs, and production etc data eventually do tend to come back to normal trends once the strike is finished. However, the lost income is gone forever, and the US economy is currently operating at a very low pace and is thus more vulnerable to ripple effects if the strike is protracted.