New Evidence on Forward Guidance Effectiveness

Fiscal policy design and the forward guidance puzzle

This paper by Ansgar Rannenberg investigates the forward guidance puzzle in Heterogeneous Agent New Keynesian (HANK) models by analyzing how different fiscal policy rules affect the transmission of nominal interest rate pegs. The author uses the Sequence-Space Jacobian method to conduct nonlinear perfect foresight simulations, comparing results across:

– A canonical HANK model with sticky wages and flexible prices
– A representative agent (RA) benchmark
– A quantitative two-asset model with capital, sticky prices and wages
– Various fiscal policy specifications

The key innovation is examining how fiscal policy design—particularly the speed and nature of fiscal adjustments to government debt—shapes forward guidance effectiveness in heterogeneous agent environments.

Findings

• Fiscal rules matter crucially: The paper demonstrates that the choice of fiscal rule dramatically alters forward guidance effects in HANK models
• Balanced budget rule: Forward guidance effects are much larger than in RA models due to redistribution from low-MPC asset holders to high-MPC households through lower taxes
• Gradual adjustment rule: Effects become much weaker than in RA models and linear in peg length, resolving the forward guidance puzzle
• Resolution of the forward guidance puzzle: Under empirically realistic fiscal rules where tax revenue responds gradually to debt deviations and economic activity, forward guidance effects become modest and linear rather than explosive. This occurs through two channels: reduced income for asset holders and higher taxes that dampen consumption increases.
• POSA approximation: A representative agent model with Preferences Over Safe Assets closely approximates HANK results under gradual adjustment rules, suggesting this simpler framework may be adequate for policy analysis.

Forward Guidance and Fiscal Rules in HANK
Author: Ansgar Rannenberg
From: National Bank of Belgium

Is forward guidance effective?

The paper by Stephen J. Cole, Enrique Martínez García, and Eric Sims examines how forward guidance effectiveness varies across time and countries when central banks interact with agents having heterogeneous expectations. The authors develop a small-scale New Keynesian model that incorporates heterogeneous expectations formation. The key innovation is allowing two types of agents:

1. Rational agents: Form fully-informed rational expectations and incorporate central bank forward guidance announcements into their forecasts
2. Boundedly rational agents: Use backward-looking forecasting models, ignoring forward guidance announcements until they materialize

The model features standard elements including habit formation, price stickiness, and an inertial Taylor rule augmented with forward guidance shocks. The aggregate expectations combine both types of agents, with the relative share determining overall forward guidance effectiveness. Finally, the authors estimate the model using Bayesian methods for the US, UK, Germany, and Japan (plus Canada, France, Italy, and Spain) from 1990Q3-2022Q3. Crucially, they use survey-based expectations data from Consensus Economics to identify forward guidance shocks and discipline the estimation.

Findings

• Cross-country variation in forward guidance effectiveness:
  – The US shows the highest share of rational agents at 47%, followed by the UK (42%) and Germany (43%)
  – Japan is a notable outlier with only 23% rational agents, indicating much lower forward guidance effectiveness
• Declining effectiveness over time:
  – All countries show substantial declines in rational agent shares over time, particularly comparing the Great Moderation (1990-2007) to the low-interest-rate period (2005-2022)
  – For example, the US share declined from 55% to 39%, the UK from 57% to 32%
  – Rolling window estimates reveal approximately monotonic decreases in forward guidance effectiveness
• Economic significance:
  – A 25 basis point forward guidance shock in the US was approximately two-thirds as effective at raising output and inflation at the end of the sample compared to the beginning
  – The declining effectiveness coincided with periods when forward guidance was used most prominently at the zero lower bound
• Policy implications: The findings suggest that central bank models typically assume full rational expectations may overestimate forward guidance potency. The reduced effectiveness over time may reflect:
  – Increased skepticism toward central bank communications
  – Greater reliance on backward-looking forecasting methods
  – The paradox that forward guidance became less effective precisely when it was used most extensively

Living Up to Expectations: The Effectiveness of Forward Guidance and Inflation Dynamics Post-Global Financial Crisis
Authors: Stephen J. Cole, Enrique Martínez García, Eric Sims
From: Marquette University, Federal Reserve Bank of Dallas