And what’s up with the University of Michigan survey measure?Figure 1: CPI inflation year-on-year (black), median expected from Survey of Professional Forecasters (blue +), median expected from Michigan Survey of Consumers (red), median from NY Fed Survey of Consumer Expectations (light green), forecast from Cleveland Fed (pink), all in %. NBER defined peak-to-trough recession dates shaded gray. Source: BLS, University of Michigan via FRED and Investing.com, Philadelphia Fed Survey of Professional Forecasters, NY Fed, Cleveland Fed and NBER.
Typically, the NY Fed and University of Michigan surveys have pretty much tracked each other (see discussion here), while economists’ forecasts and mixed survey/financial market data measures track each other and diverge from household/consumer expectations. What is unusual is Michigan’s deviation from the NY Fed’s measure in recent months.Figure 2: Median expected 1 year inflation from Michigan Survey of Consumers (red) minus median from NY Fed Survey of Consumer Expectations (light green) as of indicated date, both in %. NBER defined peak-to-trough recession dates shaded gray. Source: BLS, University of Michigan via FRED and Investing.com, NY Fed, NBER, and author’s calculations.
Now, it could be the NY Fed that has deviated, rather than the University of Michigan measure. One way to answer the question of which one deviates is to regress each measure on actual CPI inflation, and to see where there are structural breaks. I use the 1 step ahead Chow test to determine the breaks. First the NY Fed:
Now the University of Michigan survey:
The NY Fed series/Actual relationship exhibits a break associated with the inflation surge, with significance around April 2021. The Michigan series exhibits a significant break around February 2021. However, while the NY Fed series exhibits no further breaks, the Michigan series does in April 2023 and again in September 2023.More By This Author:Under Pressure: A Market Update
CFNAI, WEI, WECI, etc.
Price Level For The Carnivore’s Breakfast/Commute