Firm cost beliefs steepen Phillips curve and mute demand shocks
A Federal Reserve study by Robert Minton and Hugo Monnery reveals that US firms form cost forecasts by overreacting to past unit costs while underreacting to aggregate signals. This behavior creates a steeper New Keynesian Phillips curve, making supply shocks more inflationary than demand shocks.
Disconnected signals and past cost inertia
Analyzing survey microdata from the Atlanta Fed's Business Inflation Expectations survey between 2011 and 2024, the authors establish five key facts about firm belief formation.
A 1 percentage point shift in CPI expectations induces only a 0.15 percentage point revision in a firm's own cost forecasts, showing a major disconnect between aggregate CPI and unit cost beliefs.
Firms systematically overreact to their own past cost growth, with a perceived persistence of 0.37 against a true outcome persistence of 0.16. This overreaction remains remarkably stable across economic sectors, firm sizes, and time periods.
Moreover, firms exhibit inertia by relying on cost data from up to twelve months prior, and they underreact to public signals like crude oil prices until their own production costs actually move.
Asymmetric shock propagation and policy bounds
Calibrating an adaptive learning model yields a quarterly weight on rational expectations of just 0.25, alongside a high perceived persistence parameter of 0.82. Incorporating these microfounded cost beliefs into a standard Calvo pricing framework produces a New Keynesian Phillips curve that is steeper and less forward-looking.
Because firms only adjust prices once their own unit costs shift, supply shocks propagate rapidly into inflation.
In contrast, demand shocks generate muted inflation because firms fail to anticipate future wage pressure.
Consequently, long-horizon forward guidance loses power.
A necessary reality check
This research offers a necessary reality check to standard models assuming fully rational corporate price setters.
By linking pricing to actual unit cost histories, the authors demonstrate why long-horizon forward guidance often falls flat.
Still, assuming these behavioral parameters remain fixed during sudden inflationary shocks stays debatable.