November 1, 2023
Hard > soft
One of the most interesting phenomena in this economic cycle has been the way the soft data and sentiment releases in the US have been very poor predictors of the hard data. As such, the overlays of sentiment data that predicted calamity for the USA have not worked and that has been part of the reason that we have had this disconnect between the persistence of the recession call throughout 2023 and the much brighter reality. If most economists use soft data as a leading indicator, and it’s wrong, then most economist forecasts are going to be wrong.
There seem to be two things going on. First, the psychological and emotional damage from COVID continues to be a dark blot on the fabric of the US psyche. The pandemic blasted a hole in citizens’ trust in government, money, politics, science, and more. And it brought painful inflation. As such, consumer sentiment remains depressed, regardless of nominal economic performance. You can see this marked in yellow in the heatmap.
Second, there is this pervasive fear that higher rates will lead to economic disaster. You see it on Twitter, and you see it in business sentiment as the start of the Fed rate hike cycle cracked business sentiment and it never came back. I’m making a cause/effect statement here that is wildly simplistic, but tight money is probably at the heart of it.
Ranking economic data points vs. the prior 60 months (2019 to 2023 shown)
This sort of heatmap is a nice visual way to see the evolution of the economy over time. I took a look at the 2002-2010 cycle as well just to see if anything interesting jumps out.
The first thing I looked at was which series led as we moved into the 2008 crash. If you look at the first four columns, NFP and Claims turned first, while Continuing Claims was slow and the Unemployment Rate was very, very late. This is consistent with what you see if you overlay Claims and UR. The Unemployment Rate is a useless, badly lagging statistic. I don’t know why anyone even looks at it. Same thing every cycle.
In that cycle, NAHB obviously gave an early warning, but that cycle was more housing-centric than any other and this cycle is less housing-centric than any other when it comes to employment and economic outcomes. ISM was an OK lead indicator, in theory, except that it rebounded April/June 2007 after the Bear Stearns bailout and so again sentiment was a really bad indicator there. It just followed the Fed and the stock market, the opposite of 2022 when lower stocks and a tighter Fed dropped ISM and it meant little for the real economy.
All this shows is there is no right answer, or one size fits all methodology that works cycle-to-cycle or regime-to-regime. Some data like the US Unemployment Rate, is always lagging and never useful. But soft data is a good predictor at times, and useless at times. When sentiment is driven by politics, deaths of despair, inflation, collapsing trust in institutions, and other variables unrelated to nominal growth, it’s not a good predictor of nominal growth. That’s been the case this cycle. That’s why hard data is so much more important than soft data right now.
As such, I think the most important data points for tracking the US economy right now are:
- Retail Sales (consumer is the main driver)
- Initial Claims (timely and accurate)
- Industrial Production (hard data)
- Atlanta Fed GDPNow (better predictor of GDP than economist estimates for 10 straight quarters)
- NFP and ADP combined to reduce the noise (hard data)
What other hard data do you think will lead if and when the US economy turns weaker?
Final Thoughts
Take a look at the positioning report below. I think the broad theme of diverging nominal yield direction is interesting and makes for some compelling possibilities in the CAD and AUD crosses, for example.
Have a high-powered day.
good luck ⇅ be nimble