Asset Price Bubbles and Monetary Policy: A Multivariate Extension

1
Asset Price Bubbles and Monetary Policy A
Multivariate Extension

• By Andrew Filardo, BIS
• Prepared for a workshop on Fundamental and
  Non-fundamental Asset Price Dynamics Where Do We
  Stand?, Venastul, Norway, 14-15 February 2008

2
General MotivationGreenspan, Bubbles and Policy
In conclusion, the endeavors of policy makers to
stabilize our economies require a functioning
model of the way our economies work.
Increasingly, it appears that this model needs to
embody movements in equity premiums and the
development of bubbles if it is to explain
history. Jackson Hole 2002
3
Key Contributions Of This Paper

• Modelling endogenous multivariate bubbles in a
  dynamic macro model

• Methodology relatively simple macro setup and
  realistic multivariate bubble specification do
  work and yield key insights

• Monetary policy conventional wisdom about
  benign neglect and mopping-up later approach
  may be too optimistic in the current policy
  environment

4
Preview Of Findings

• It is optimal to respond to asset prices
  generally and bubbles specifically!

• Multivariate bubble considerations complicate the
  policy tradeoffs

Defensive strategies preventing and pricking
asset price bubbles Opportunistic strategies
use bubbles to achieve stabilization goals
5
Monetary Policy And Asset Prices

• Asset price booms and busts have been extreme
  developments that monetary authorities have had
  to face.

6
Monetary Policy and Asset Prices
Equity
Equity
Housing
Housing
7
Monetary Policy and Asset Prices Asia-Pacific
Housing
Equity
8
Monetary Policy an Asset Prices

• Asset price booms and busts have been extreme
  developments that monetary authorities have had
  to face.

• Asset price booms and busts are an important
  feature of the monetary policy landscape going
  forward. Borio, English and Filardo (2003)
• The double-bubble aspect of this relationship
  has been underappreciated

9
Road Map

• Lay out monetary policy model
• Multivariate bubble specification
• Results
• Policy implications and conclusions

10
The Greenspan Approach Conventional Wisdom?
Greenspans preferred approach to bubbles is
to let them burst of their own accord, and then
to use monetary policy (and other instruments),
as necessary, to protect the banking system and
the economy from the falloutthe mop-up after
strategy If the mopping up strategy worked
this well after the mega-bubble burst in 2000,
shouldnt we assume that it will also work after
other, presumably smaller, bubbles burst in the
future? Blinder Reis,
Jackson Hole 2006
11
Three Modelling BlocksSmall-Scale Macro Model
12
Asset Price Block
13
Monetary Policy Block
Standard Loss Function
14
Optimal Monetary Policy
subject to the model of the macroeconomy and
asset prices
15
3 Policy Specifications
16
Modelling the Bubble
Time-Varying Transition Probability (TVTP) Model
For Each Bubble
17
Modelling the Bubble
Time-Varying Transition Probability Model
Sample Path of a Bubble - Blowing Bubbles
15
10
5
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
-5
-10
-15
-20
Time periods
18
Modelling the Bubble
No-Bubble State Transition Probability
Function of yt-1
19
Modelling the Bubble
Bubble State Transition Probability

• 2 specifications
• - weakly interacting multivariate bubbles
• - strongly interacting multivariate bubbles

Housing bubble transition probability
Own duration dependence
for strong version
20
Results for Optimal Policy
Optimal Monetary Policy Frontiers
Variance of inflation
Variance of output
21
Results for Optimal Policy
Optimal Monetary Policy Frontiers
Variance of inflation
Superior policy
Variance of output
22
Results for Optimal Policy
Optimal Monetary Policy Frontiers
Variance of inflation
Variance of output
23
Result for Standard Taylor Rule
Optimal Policy Frontiers
No response to AP
24
Basic Result I Pricking AP Is Optimal
Optimal Policy Frontiers
Response to AP
No response to AP
25
Basic Result 2 Knowing NF and F not critical
Optimal Policy Frontiers
Response to AP
No response to AP
Response to F and NF
26
A Little Bit of Nonlinearity, Important
Consequences
Bubble part
27
Complex Dynamics in Model
Output
28
Nonlinear Impulse Responses
Inflation
29
Nonlinear Impulse Responses Illustrate

• Endogenous bubble models admit richer dynamics
• Chaotic behavior is possible
• Monetary policy helps to stabilize the economy
• by not only pricking bubbles but also by
  exploiting bubbles (Blanchard, 2000)

30
Expected Durations

• How do the expected durations of a bubble change
  over time?
• - Assume in the no-bubble steady state and a
  pickup in economic activity begins
• - In simulations, an 8-period positive shock to y

31
Expected Durations Monetary Policy Matters!
32
Expected Duration for Housing Bubble
33
Expected Duration for Housing Bubble
Counterfactual simulation (?)

• Start with an equity bubble (1995)
• Then a housing bubble begins (1999)
• Then recessionary shocks start (2000) soon
  after the equity price bubble collapses
• Subpar growth shocks end (2002)
• Equity price bubble restarts (2003)

34
Expected Duration for Housing Bubble
Recession ends at point D and optimal R increases!
Recession
E bubble
H bubble
35
Expected Duration for Housing Bubble
Too Low, Too Long Policy
36
Summary So Far

• Optimal policy indicates central banks should
  respond to asset price bubbles on the way up as
  on the way down ? in other words, no Greenspan
  benign neglect on the upside
• Also, double bubbles are a double whammy on
  Greenspan approach ? dont be too aggressive on
  the downside
• Consensus view a la Blinder/Reis not robust!

37
Extra Consideration Closer to Reality
Incorporating Uncertainty Into Analysis
Compare the expected gains and losses from
reacting to asset prices Risk Management!
38
Bayesian Expected Loss Analysis
39
Implications for Monetary Policy and Benign
Neglect
Degree of confidence in bubble matters!
Threshold
40
Conclusions
In this class of models, generally monetary
authorities want to respond to asset prices it
also optimal to prevent and prick them.
Findings about optimal monetary policy in
multi-bubble environment - Intuition from
single-bubble models might be misleading at
this policy juncture - Greenspan approach in
2000 may not be as attractive in 2008
41
Conclusions
Whats next on policy modelling front? -
Forward-looking behavior (lengthening
horizon!) - Uncertainty with small
probability events (tail risk, insurance
motives and Greenspan/Bernanke put) -
Optimal policy mix especially at the ZLB
42
Conclusions
Whats next on bubble modelling front?
Better micro-foundations for financial
imbalances and bubbles - What is the role of
money and credit? - What is the nature of the
coordination failure? - What is the role of
rational herding and information cascades?
Other behavioral models of asset prices?
43
Where do we stand?... Now, In Disarray
Investors expect the federal funds rate to be as
low as 2.25 by the end of the year...In trying
to prevent financial-market calamity, the
(Bernanke) Fed may find itself pushed by Wall
Street to leave interest rates too low for too
long. The Economist, 31 January 2008
So, heres an alternative explanation for why the
timing and size of the 75 basis point
intermeeting cut in the federal funds rate
target. Chairman Bernanke and his colleagues want
us to know that when they see changes in the
economy that compromise their medium-term
stabilization objectives, they will act. January
23, ???
44
Thank you