For all the talk of a ‘soft landing’ or how buoyant and well-positioned the consumer is, there are indications of weakness in the consumer sector in the US
- Looking at snapshots of US car auction data, and restaurant and retail spending, we have identified an age cohort who are seemingly beginning to feel the pinch, with certain purchases being delayed
- The question is, will the weakening of spend for this age cohort continue, and are the conditions there for other age cohorts to follow?
When it comes to credit investing, you’re allowed to be early – but you can’t be wrong. At the beginning of this year there was an expectation of recession. We all know how wrong that call was. Instead, a day doesn’t go by when we don’t hear about an impending “soft landing”, or how buoyant and well-positioned the consumer is, or why despite sentiment numbers being horrific “things really aren’t actually so bad”.
However, there are indications of weakness. Apparel and discretionary goods sales are bad, and not just among the more cash-strapped US consumer. Private labels are consistently taking share from branded packaged food for the first time since the Covid-19 pandemic began. Credit usage is growing quickly1, as are delinquencies. So, what gives – are we on the brink of a stressed consumer and a subsequent strangle of discretionary spending or not?
Our proprietary research suggests that, under the surface, the 25-34-year-old cohort of consumers are cutting back on discretionary spending quickly, and the robust levels of discretionary spend broadly across cohorts is more delicate than it seems.
Unique dataset
The recent explosion in popularity of online US car auctions has created a unique two-way market dataset. The market is liquid, nationwide, and has a significant amount of data – more than 10,000 completed auctions this year – with new results daily. Using our proprietary scraping tool, we have captured and catalogued auction results over the past five years for the BMW M3. For those unfamiliar with it, the M3 is BMW’s motorsports-derived version of its legendary 3 Series – arguably the source of BMW’s classic tagline “The Ultimate Driving Machine” – and has achieved legendary status among enthusiasts. We have compared four generations of BMW M3, each representing a different cohort of collector and enthusiast. Our hope was that some divergences in the data might be predictive in giving us a clue towards coming stress.
It would be remiss to not caveat this work with the limited nature of the dataset. Because online car auctions are only a recent phenomenon, five years is as long as we can go back. For that reason, this should be taken as an incremental input, not an absolute driver of any broad conclusion.
So, what did we find?
We learned that the best way to identify divergences in the dataset was to plot the spread of values for different generations of M3. When we cleaned the data it became apparent that the E36 (manufactured from 1992-1999) and the E46 (2000-2006) had the most results, were the most liquid markets, and had the strongest relationship. We therefore focused our analysis on the relationship between those two generations, seeking to ultimately identify if there was a point at which the spread quickly widened or tightened beyond the historical range. As Figure 1 shows, at the beginning of 2023 the spread widened steeply beyond its five-year wides and hasn’t stopped, which indicates that the value of enthusiast-grade E46 M3s is dropping much more quickly than those of the previous generation.
Figure 1: BMW E36 and E46 price spreads (%)
Source: Columbia Threadneedle Investments’ analysis, October 2023
When you examine real retail spending (adjusted for inflation) by age cohort (Figure 2) it is clear that the different age groups directionally move together for the most part. This makes sense – discretionary spending historically tends to be similar directionally across cohorts. Put another way: even rich people cut back when things get bad. But look at March 2023: where the other age cohorts’ retail spending rebounded, the 25-34-year-olds appear to continue declining in real terms. The M3 Index spread preceded that divergence by three months.
Figure 2: real retail spending, percentage change
Source: Federal Reserve Bank of New York, June 2023
Also notable in this dataset is that retail spending for the oldest cohort, who are also the wealthiest, is unchanged versus the end of 2020. It was the 25-34-year-old cohort that increased spending the most, and therefore has the furthest way to fall.
The next step was to see if there were any other similar datasets that could give us some insight into the stronger parts of the consumer economy – like services and hospitality – given the “services over goods” narrative. Here, the 25-34-year-old cohort is also clearly spending less at restaurants (Figure 3), diverging with the other rebounding age cohorts.
Figure 3: real restaurant spending by age, percentage change
Source: Federal Reserve Bank of New York, June 2023
The M3 Index spread preceded that divergence by two months. This is the same dynamic that was apparent in the retail spend data. Interestingly, however, that trend is not reflected when the data is split only by income level (Figure 4).
Figure 4: real restaurant spending by income, percentage change
Source: Federal Reserve Bank of New York, June 2023
Conclusion
The spread divergence in our M3 Index preceded the divergence of 25-34-year-old discretionary spending versus other cohorts. Our theory is that the E46 M3 was the “it” car when enthusiasts who are now 30-35-years-old became legal drivers. That is the car those enthusiasts wanted when they had some expendable income, and it was the first to be delayed when it came to halting any unnecessary purchases given its expense compared to buying new clothes or eating out at a restaurant.
The questions we must ask are: 1) will the weakening of spend for this age cohort continue? And 2) if it does, will other age cohorts follow?
The recipe is there for a rapid broad deceleration of consumer spending and all it may take is a modest acceleration of the unemployment rate to trigger it.
