The necessary and sufficient conditions for Fed tightening within our 2019 #FloorGLI framework remain satisfied, but the goals shouldn’t be to increase unemployment and the case for rapid 75bp rate hikes is diminishing…
In May 2019, we released our Fed framework. Prioritizes a robust trajectory for gross labor income (GLI) (job growth & wage growth).
If expecting GLI to fall short of "floor" growth trajectory. ease.
If expecting above floor & inflation > target, tighten employamerica.org/researchreport…
In January 2020, we offered a more operational revision to the framework. National accounts measures of labor income are heavily revised; suboptimal for real-time policymaking. CPS employment and Employment Cost Index are better suited (no revision issues) medium.com/@skanda_97974/…
Threshold for revising/deviating from any pre-specified framework should be high; no strong reason to do so here.
Income growth remains strong, and inflation pressures remain. Given that, Fed is justified (within limits) to tighten financial cond'ns for sake of lower inflation
Moreover, labor income levels already "caught up" to the pre-pandemic "floor" in 2022Q1 itself.
We specified in 2019 that if the Fed was constrained from conventional easing (@ Zero Lower Bound) & incomes were falling short of "floor", condition liftoff on catch-up
Where we most firmly depart from Fed is on their perverse 'optimal policy' projections: Engineering recessionary unemployment increases involve more persistent economic damage.
GLI growth (job growth & wage growth) can slow w/o resorting to sharply higher unemployment rates
We noted this point in our piece "Beyond The Phillips Curve":
Declining labor utilization rates tend to predict *further declines.*
The exceptions to this rule of thumb tend to involve policy interventions to actively stabilize employment
The persistence of labor utilization declines is also not the kind of easily reversible phenomenon that textbook might suggest. Prime-age male employment systematically fails to recover from recessionary declines (while surging inflation can revert w/o recessions)
At tomorrow's Fed meeting, it is all but given that the Fed will be hiking 75bps. While the direction of tightening can be justified under current macroeconomic conditions, the case for such rapid tightening is diminishing.
Rate hikes are not a neatly linear policy tool
Two countervailing dynamics that raise uncertainty around "how much much more to tighten" (and elude "r-star" models): 1. Financial conditions tightening >>> Fed hikes 2. Real economy ex-housing is showing less sensitivity to rate hikes
#2 could just be a f'n of longer lag
Our preferred framework endorses a "change rule" approach to setting rates; celebrated r-star estimates are heavily backward-looking.
Better to take it meeting by meeting, see how macro cond'ns evolve, and support acceleration/deceleration accordingly
The macro conditions tell us that we're still seeing an outlook for strong income growth along with high inflation. But also, income growth is more clearly decelerating, both in terms of job growth and wage growth (likely for its own reasons, not b/c Fed)
Having won back the key jobs lost to the pandemic (e.g. see chart), the case for slowing labor income growth from elevated rates is not as costly now. But given that deceleration is underway, financial conditions tightening so much already, & l/r costs of overtightening high...
the case is strengthening for taking it slower on its tightening efforts.
This wouldn't foreclose where it might ultimately take its policy rate, but @ current pace, Fed isnt giving itself much time to evaluate economic effects from realized financial conditions tightening /END
• • •
Missing some Tweet in this thread? You can try to
force a refresh
This paper immediately made the rounds b/c it got all 50 states right w/ a model based on "fundamentals"
Kudos to them on the favorable prediction
But I think some lay readers might miss that "fundamentals" encompasses more than economic data
Worth digging a little deeper 🧵
The fundamental economic data they rely on is the Philly Fed Coincident Index, an composite of labor market, manufacturing, and real income data available at the national and state-level data
Here's what the year-over-year change in the Index looks like
The authors use a weighted average of the monthly readings through the presidential term. Likely from the month of Inauguration to the June of the election.
Their weights are in an unpublished appendix.
So I followed the cited approach to replicate their model input
Payrolls: Great! But more a sign of lower layoffs in Jan than a hiring uptick (seasonal adj issues here)
Wages: Still solid real wage gains but not spike reflects how weather effects on workweek distort data
Prime age employment: Nice bounce but wanna see more soon
False start...now we're up and running
A nice bounce in the prime-age employment rate, which already adjusts for changes in (1) participation and (2) aging
But even so, it doesn't reverse December's decline. Still 0.3% below peak vs July
Still looks like a slowing labor market
The household survey looks weaker when we look at the subset of full-time employment, which naturally controls for part-time underemployment issues. Another decline, now down YoY
No labor market statistic has a monopoly on the truth. Good to take an avg across hot and cold data.
We have 6-7 months of solid core disinflation to point to. It's fantastic, but there are still potential bumps in the road. Let's get a clean 12 months in first, especially given residual seasonality.
Not yet time for decisive retrospectives
1. If people can’t agree on what everyone else meant by “transitory,” the debate is futile Healthy debate requires mutual intelligibility and charity.
Instead, there's plenty of confusion about what each person meant explicitly, leave aside implicitly
Today's numbers aren't a basis for outright pessimism but they do signal the need for caution
Even looking beyond distortions due to the UAW strike, trends in job/wages/hours are cooler across both surveys
Fed needs to take notice & avoid overtightening
Nonfarm payroll employment is most frequently cited for job growth:
October readings were held down by the UAW strike, but the hidden story was the downward revision to the previous two months' jobs numbers.
Underlying job growth is still solid in the survey but it's slowing
The household survey shows weaker job growth
Prime-age employment rates (which acct for participation, aging) declined again. Full-time also weaker
Similar decline occurred last fall (before reverting). Could just be bad seasonal adjustment but warrants elevated caution rn
1.1% through lower costs as per its specific PPI deflator
Construction costs blew out in 2021-22 for a variety of reasons (hot housing market, covid-zero affecting supply chains, Olympic blue affecting building material supply, etc). All of that depressed business construction capex
Supply chain healing now helping the trajectory here
A 🧵 on how everyone who cites unit labor costs as a predictor / explanation for inflation misses the fatal near-tautological error they’re committing (tbc not aimed @KathyJones, who you all should follow).
ULC is supposed reflect “wages divided by productivity” but…
Unit labor costs do not involve any direct measure of an individual business’ unit labor cost. Nor does it involve direct measures of a firm’s wages or productivity. It comes from aggregates
= (Total Labor Comp / Total Hrs Worked) / (Total Real Output / Total Hrs Worked)
Total Hrs Worked drops out of numerator & denominator
ULC = Total Labor Comp / Total Real Output
Slight tracking error due to “nonfarm” concept and labor comp definition when applied to ULC estimate but you should get the idea from this chart overlay