I have a secret and a confession.
Below the surface of embraced uncertainty, beyond the deep recesses of objectivity and through the event horizon of data agnosticism, there sits a sheltered pool of stealth bias.
While I often have a convicted view on the probable slope of the line for any given high-frequency domestic macro series, I occasionally harbor a desire for a specific outcome.
This morning offers a quintessential example as I was hoping Retail Sales would print pretty much exactly the way it did.
For 2 reasons:
- Expositional convenience: It offers an easy case-study in Macro modeling 101.
- 3Q/2H Trajectory: It sits as a microcosm for how the macro data is likely to evolve over the coming quarter(s)
With respect to #1:
- Isolated distortions and large-scale macro shocks muddy the comp waters. If something was down -100% last year due to some transient shock, of course it’s going to be up big in rate-of-change terms in the corresponding period the following year….
- The analytical question of consequence is: Outside of base effect impacts, is the underlying trend actually improving or deteriorating?
- The primary, lowest intensity means to normalize against transitory distortions is to use the 2Y growth comps. For example, for June 2021, you would just take the average Y/Y growth in June 2021 and June 2020.
With Respect to #2:
- What I expect to happen, particularly across the Services Sector, is for the Y/Y comps to peak and fade (that has begun) but for the 2Y comps to be flat-to-accelerating … the signal being that the underlying trend continues to improve. This is what we observe in this morning’s Retail Sales data.
- There also exist nested distortions (distortions within the distortions). For example, Mar-May of 2020 marked the trough COVID comps, so Mar-May 2021 were destined for RoC solidity … but we also got another round of stimulus checks that hit in March ’21 … with that income serving to amplify consumption in March and April. You then got a little bit of sequential hangover in May as that stimulus impulse came off. So what does that mean for a little ‘follow the 2Y comp’ exercise? …
- Basically, it means that while the underlying trend remains one of ongoing improvement, the 2Y comp won’t be totally linear. It should (and did) peak alongside the layered distortion in March/April …. Decelerate in May, then continue its upward trajectory. Which it did – see 1st chart below.
- Again, this temporal comp procession will be more pronounced in Services as consumption progressively renormalizes away from the pandemic amplified spending on Goods.
Understanding the underlying trend will be key in traversing the chop associated with a developing macro regime shift. It will also offer explanatory power in the two-way communication loop with our Quantitative/Risk Management Signaling process.
Coming off the easiest comps in modernity, we will get a mechanical deceleration in growth/inflation with a trajectory that puts on course for Quad3/Quad 4. While some key Trend signals have already flipped in the direction of Quad 3, Quad 2 had not (fully) ceded the drivers seat with respect to performance.
This would make sense if anomalous macro conditions were driving decelerating Y/Y growth but the underlying trend (normalized for comp dynamics) was still improving. To the extent the Y/Y and 2Y growth numbers are telling divergent stories I expect elevated market consternation around the appropriate asset/sector/style factor allocation.
I suspect it will get more interesting and the breadth/strength of the signal from the Signal (so to speak) will become more pronounced should both the 1Y and 2Y growth numbers start to flag simultaneously and conspicuously.