Discussion of
“Forward Guidance by Inflation-Targeting Central Banks”
By Michael Woodford
Sveriges Riksbank Conference
“Two Decades of Inflation Targeting: Main Lessons and Future Challenges”
June 3, 2013
John Williams
Federal Reserve Bank of San Francisco
The opinions expressed here are those of the author and do not necessarily reflect the views of
anyone else in the Federal Reserve System.
The Ascent of Inflation Targeting
FIT with policy
Roadmap of Discussion
• Competing Visions of Inflation Targeting
– “Classic” IT vs. Woodford’s Ideal IT
• Forward Guidance
– Experience of the Federal Reserve
• Nominal GDP Targeting
– Theory and Practice
Competing Visions of IT
• “Classic” Inflation Targeting
– Conservative: limited goals
– Primacy of inflation objective
– Communication focused on anchoring inflation
– Avoidance of policy errors of past
• Woodford’s Ideal IT
– Ambitious: implement & communicate optimal
policy approach
– Fine-tuning of policy and communication
“Classic” Inflation Targeting (RBNZ)
Woodford’s Criteria for Ideal IT
• Implements optimal policy under
commitment (yielding unique RE equilibrium)
• Communicated in terms of a criterion of
forecasts of target variables
• Performs well in presence of zero lower
bound on nominal interest rates
• Robust to forecast/policy errors
• Conclusion: variant of price level targeting or
nominal GDP targeting preferred
Forward Guidance
• The Federal Reserve’s use of forward
guidance has evolved dramatically over the
past decade
– Policy statements now make explicit reference to
economic conditions for policy action (for funds
rate liftoff and continuation of asset purchases)
– FOMC quarterly projections of the funds rate
• Forward guidance has clearly affected
market expectations of future path of rates
(Swanson and Williams 2012)
Effectiveness of Forward Guidance:
Event Studies
See also: Bernanke-Reinhart-Sack (2004), Kohn-Sack
(2004), Gürkaynak-Sack-Swanson (2005), CampbellEvans-Fisher-Justiniano (2012)
Effectiveness of Forward Guidance:
Shift in Private Forecasts
Effectiveness of Forward Guidance:
Reduced Uncertainty based on Options Data
Probability that U.S. funds
rate < 50bp in 5 quarters
Probability that UK
Libor rate < 75bp
in 12 months
Explicit forward
Guidance (US)
Forward Guidance: Practical Constraints (I)
• Lack of strong theoretical foundations and
empirical evidence on what constitutes “best
practice” CB communication.
– Why is CB communication necessary? What is source
of breakdown of rational expectations equilibrium?
– How does one communicate intentions vs.
commitment, uncertainty and disagreement?
• Public’s “rational inattention” creates limits and
unavoidable tradeoffs for CB communication:
– Central bank communication is allocated only so much
bandwidth, which calls for simplified clarity on a few
points, rather than comprehensive description of policy
– Example: FOMC 6½ unemployment rate threshold
Forward Guidance: Practical Constraints (II)
• Diversity of views makes it difficult to agree on and
describe “consensus” policy reaction function (or
targeting criterion); instead, FOMC communicates
through multiple channels:
– consensus statement on longer-run goals and policy
– FOMC policy statements and minutes
– quarterly policy projections based on individual
assessments of appropriate monetary policy
– speeches and testimony
• As a result, much of communication centers on turning
points such as rate liftoff and end of asset purchases,
rather than more general policy reaction function.
• At times, multiplicity of communication channels can
interfere with signal clarity
Nominal GDP Targeting: Theory
• NGDP Targeting has advantages of PLT:
– Approximates optimal policy
– Robust to natural rate and model uncertainty
(Orphanides and Williams 2002, 2006)
– Performs well in presence of zero lower bound if
credible and understood by public (Williams 2006)
• Imposes “balanced approach” to objectives ( =1 )
• Risk sharing of nominal debt contracts
– Koenig (IJCB 2013)
• Relatively easy to communicate to public (?)
Nominal GDP Targeting: Theory
• NGDP targeting may be robust to high
degree of natural rate and model uncertainty
• Example: robust difference rule (OW 2006)
i = 1.1(π-π*)− 2.6 u
1.1{ y + π - ( y* + π*)}
i = 1.1{y + p - (y* + p*)} + const.
u: unemployment rate
y: real GDP (u & y linked by Okun’s Law)
Nominal GDP Targeting: Practice
• Unanticipated shifts in potential output growth
translate into persistent deviations of inflation from
desired target
• Policy should be symmetric (example: what would
PLT or NGDP targeting imply for the UK today?)
• Relies on credible commitment to not let “bygones
be bygones” which may be difficult to
communicate or implement in practice.
– For example, if PLT had been implemented in US in
2004, policy would be leaning against a price level
gap because of the bump in inflation over 2005-08.
Revisions to CBO’s Estimates of Potential GDP
Real GDP
Billions of
2007 Vintage
Potential GDP
2013 Vintage
Potential GDP
2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
Source: Bureau of Economic Analysis, CBO
Price Level Base Year Matters
U.S. GDP Price Level Gap (2% trend growth)
2004 Base Year
2007 Base Year
How Big is the Nominal GDP Gap?
Output gap
Price level gap
CBO 2007
( -11% )
2007 base year
( - 2% )
- 13%
CBO 2013
( - 5½%)
2004 base year
( + 1% )
- 4½%
CBO 2013
( - 5½%)
Inflation gap
- 1%
Policy gap
(equal weights)
- 6½%
• If nominal GDP target based on 2007 estimate of trend
nominal GDP (up to now), monetary policy would need to
engineer a 7½ percentage point increase in price level.
• For example, that would imply 3½ percent inflation on
average over the next 5 years, which would be hard to
explain and could unmoor inflation expectations.

similar documents