Download Slides 1

Uncertainties in Monetary
Policy Design
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Aksoy, Y, and T. Piskorski (2006), US Domestic Money, Inflation and Output, Journal of Monetary Economics, 53,
pp.183-197.
Aruoba, B. (2008), Data Revisions are not Well-behaved, JMCB.
Bernanke, B. S. and J. Boivin, “Monetary Policy in a Data-Rich Environment,” Journal of Monetary Economics, 50,
April 2003, 525-546.
Cayen, Jean-Philippe & van Norden, Simon, 2005. "The reliability of Canadian output-gap estimates," The
North American Journal of Economics and Finance, 16(3), pp. 373-393.
Croushore, D. and C. Evans, (2006), Data Revisions and the Identification of Monetary Policy Shocks, Journal of
Monetary Economics, 1135-60.
Croushore, D. and T. Stark, (2003), A Real Time Dataset for Macroeconomists: Does the Data Vintage Matter?
Review of Economics and Statistics, pp. 605-17.
Faust, J., Rogers J.H. and Wright, J., (2005), News and Noise in G-7 GDP Announcements, Journal of Money,
Credit, and Banking, 37, pp. 403-19.
Faust, J., Rogers J.H. and Wright, J., (2003), Exchange Rate Forecasting: The Errors We've Really Made, Journal
of International Economics, 60, pp. 35-59.
Mankiw, N.G. and Shapiro, M.D. (1987), News or Noise? An Analysis of GNP Revisions, NBER Working Paper
No. 1339, Cambridge MA.
Mankiw, N.G., Runkle, D.E. and Shapiro, M.D. (1984) Are Preliminary Announcements of the Money Stock
Rational Forecasts? Journal of Monetary Economics 14, pp. 15-27.
Orphanides, A., (2001), Monetary Policy Rules Based on Real-Time Data, American Economic Review, 91,
pp. 964-985.
Orphanides, A., (2003), Monetary Policy Evaluation with Noisy Information, Journal of Monetary Economics, 50,
pp. 605-631
Orphanides, A. and van Norden, S., (2002), The Unreliability of Output Gap Estimates in Real Time, Review of
Economics and Statistics, 84(4), pp. 569-583.
Orphanides, A., (2000), The quest for prosperity without inflation, Journal of Monetary Economics, 50(3), 633-663.
Debate
• CB should minimize inflation volatility and the
volatility of the gap between output and the
flexible-price equilibrium level of output.
• debate on the best strategies for achieving these
goals.
– optimal policies versus simple instrument rules
(Taylor (1993)).
– Policy based on inflation forecast targeting, nominal
income growth, price level targeting, exchange rate
targeting etc.
– Financial crisis
Current Consensus and Debate
– CB is assumed to know the true model of the
economy
– CB observes accurately all relevant variables.
– CB knows sources and properties of
economic disturbances
– Uncertainty arises only due to the unknown
future realizations of these disturbances.
– OR…..
structural change and uncertainty
• In practice tremendous uncertainty about
the true structure of the economy, the
impact policy actions have on the
economy, and even about the state of the
economy.
November 2007 CPI Fan Chart
Copyright © Pearson
Education, Inc. and
Slide 7
November 2010 CPI Fan Chart
August 2012 CPI Fan Chart
November 2007 GDP Fan Chart
Current GDP projection based on market interest rate expectations
The fan chart depicts the probability of various outcomes for GDP growth in the future. If economic circumstances identical to today’s were to prevail on 100 occasions, the MPC’s best collective judgement is that GDP
growth over the subsequent three years would lie within the darkest central band on only 10 of those occasions. The fan chart is constructed so that outturns of GDP growth are also expected to lie within each pair of the
lighter green areas on 10 occasions. Consequently, GDP growth is expected to lie somewhere within the entire fan chart on 90 out of 100 occasions. The bands widen as the time horizon is extended, indicating the
increasing uncertainty about outcomes. See the box on pages 48–49 of the May 2002 Inflation Report for a fuller description of the fan chart and what it represents. The dashed line is drawn at the two-year point.
Copyright © Pearson
Education, Inc. and
Slide 10
November 2010 and 2011 GDP Fan Charts
August 2012 GDP
Very Broadly: Three Types
• Data Uncertainty
• Parameter Uncertainty
• Model Uncertainty
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
Percent
Real Output Growth for 1977Q1
(as viewed from the perspective of 113 different vintages)
10
9
8
7
6
5
4
Vintage
Data Uncertainty
• Economic agents (policymakers, financial
agents, firms, households) possess information
and form their forecast in real time
• Most macroeconomic data is subject to
continuous revisions
• Lomax (2004), ‘few subjects consume more of
[the Monetary Policy Committee’s] time and
energy’.
First source of uncertainty: Data
– Short run revisions based on additional
source data
– Benchmark revisions based on structural
changes or updating base year
What Do Different Types of Data
Revisions Look Like?
ONS Case
– The Office for National Statistics (ONS)
publishes early estimates based on the
survey responses available at the time.
– These estimates are inevitably revised
• as more information is received.
• as statistical methods change.
• To ensure comparability of the National Accounts
through time, the ONS reconsiders the back data
in the light of any methodological changes —
leading to further revisions.
Data Uncertainty
• Preliminary data: consists of the first reported
date for each variable at each point in time
• Real time data: full vector of observations at
each point in time for each variable (maybe
seasonally adjusted or not adjusted)
• Final data: data obtained after ‘successful
revisions’ are made and no further revisions are
needed (does it exist?)
04q1 04q2 04q3 04q4
RT
03q4
04q1
04q2
04q3
y*
y
y
y*
y*
y
y*
y*
y*
y***
Final
BoE Quarterly Bulletin 2007 Q3
Revisions
X t  X t  rt
f
p
f
– Xpt denote a statistical agency’s initial
announcement (say at t+1) of a variable that was
realized at time t
– Xft denote the final or true value of the same
variable.
– rft is the final revision which can potentially be
never observed.
Real Time Data: An example
• Federal Reserve Bank of Philadelphia
http://www.phil.frb.org/econ/forecast/reaindex.html
• CB bulletins
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
Percent
Real Output Growth for 1977Q1
(as viewed from the perspective of 113 different vintages)
10
9
8
7
6
5
4
Vintage
Aruoba (2008) on US
Macroeconomic Series
• Continuous revisions
• revisions do not have a zero mean, which indicates that
the initial announcements by statistical agencies are
biased.
• revisions are quite large compared to the original
variables and they are predictable using the information
set at the time of the initial announcement therefore
– initial announcements of statistical agencies are not rational
forecasts.
– evidence that professional forecasters ignore this predictability.
Revisions
X t  X t  rt
f
p
f
– Xpt denote a statistical agency’s initial
announcement of a variable that was realized at
time t
– Xft denote the final or true value of the same
variable.
– rft is the final revision which can potentially be never
observed.
Well behaved revisions! Three
properties
• mean zero. i.e. initial announcement of the
statistical agency is an unbiased estimate
of the final value.
• variance of the final revision to be small,
compared to the variance of the final
value.
• final revision to be unpredictable given the
information set at the time of the initial
announcement.
i.e.
E (r )  0
f
t
f
var(r )
t
small
E (r I )  0
f
t
t 1
Unconditional properties of data
revisions (Aruoba (2008)
BoE Quarterly Bulletin 2007 Q3
Faust et al. (2005)
Data Revisions
• Are Data Revisions News or Noise?
– Data Revisions Contain News: Data are
optimal forecasts, so revisions are orthogonal
to early data; revisions are not forecastable
– Data Revisions Reduce Noise: Data are
measured with error, so revisions are
orthogonal to final data; revisions are
forecastable
Forecastability of Final Revisions:
Noise versus News
• Noise: the initial announcement is an
observation of the final series measured with
error. (revision is uncorrelated with the final
value but correlated with the data when the
estimate is made)
• News: the initial announcement is an efficient
forecast that reflects all available information
and subsequent estimates reduce FE! (revision
is correlated with the final value but uncorrelated
with the data when the estimate is made, i.e.
unpredictable with using the information set at
the time of the initial announcement)
Forecast efficiency tests (news vs.
noise)
• Preliminary data
X tp  X t f  rt f
under noise
rt f  X t f
but rt f correlated w. X tp
under news
rt  X t
f
p
f
but rt correlated w. X t
f
Mincer&Zarnowitz forecast
efficiency test
• If et is correlated with Xtp (noise) it will help
to predict the subsequent revision.
• Define
R(t )  X t  X t
f
p
regress
R t     X t  ut
p
forecast efficiency implies     0
Faust et al.
Data Revisions
• Are Data Revisions News or Noise?
– Mankiw-Runkle-Shapiro (JME 1984): Money data
revisions reduce noise
– Mankiw-Shapiro (SCB 1986): GDP data revisions
contain news
– Faust et al. (2004) G7 real GDP
Italy, Japan, UK about half the variability of
subsequent revisions are predictable
US: very slight predictability
– Aruoba (2008) mostly noise (except unemployment
and capacity utilization)
Data Revisions
• Is the Government Using Information Efficiently?
• Should the government report its sample
information or project an unbiased estimate
using extraneous information?
Data Revisions
• Are Revisions Forecastable?
– Key Issue: are revisions forecastable in realtime (or just ex-post)?
• Faust-Rogers-Wright (JMCB, 2005): Examines G7 countries’ output forecasts; find Italy, Japan &
U.K. output revisions forecastable in real time
(news) US revisions very slightly predictable (likely
noise)
• Aruoba (2008) similar finding, documents also that
this forecastability was never exploited
Forecasting
• Does the fact that data are revised matter
significantly (in an economic sense) for
forecasts?
Forecasting
• Leading indicators: seem to predict
recessions quite well.
• But did they do so in real time?
• Diebold and Rudebusch (1991): real-time
data
• The leading indicators do not lead and
they do not indicate!
date
1974:08
1974:07
1974:06
1974:05
1974:04
1974:03
1974:02
1974:01
1973:12
1973:11
1973:10
1973:09
1973:08
1973:07
1973:06
1973:05
1973:04
1973:03
1973:02
1973:01
Value of Leading Index
Leading Indicators, vintage Sept 1974
185
180
175
170
165
160
155
150
145
140
date
1974:08
1974:07
1974:06
1974:05
1974:04
1974:03
1974:02
1974:01
1973:12
1973:11
1973:10
1973:09
1973:08
1973:07
1973:06
160
1973:05
1973:04
1973:03
1973:02
1973:01
Leading Index
Leading Indicators, vintage Sept 1974 and Dec. 1989
185
100
180
98
Dec. 1989 vintage
175
96
170
94
165
92
Sept. 1974 vintage
90
155
88
150
86
145
84
140
82
Forecasting
• Optimal Forecasting When Data Are
Subject to Revision
– Summary: There are sometimes gains to
accounting for data revisions; but
predictability of revisions (today for US data)
is small relative to forecast error (mainly
seasonal adjustment)
Monetary Policy: Data Revisions
• How Misleading Is Monetary Policy
Analysis Based on Final Data Instead of
Real-Time Data? (Orphanides)
• How Should Monetary Policymakers
Handle Data Uncertainty? (News/noise)
Monetary Policy: Data Revisions
• How Misleading Is Monetary Policy
Analysis Based on Final Data Instead of
Real-Time Data?
– Croushore-Evans (JME, 2006): Data
revisions do not significantly affect measures
of monetary policy shocks
– Examples based on the VAR evidence a la
CEE (1996) and Gali (1992).
– Orphanides (2003) striking analysis!
Identifying the Monetary Policy
Shocks
• CEE (1996)
• recursively identified VAR in six variables:
–
–
–
–
–
–
real GNP (or GDP) (Y),
implicit GNP (or GDP) deflator (P),
non-borrowed reserves (NBR),
federal funds rate (FF),
total reserves (TR), and
an index of commodity prices (PCOM), all in logs.
• Using the Choleski decomposition, the causal ordering of
the variables is important
• use the CEE benchmark ordering Y, P, PCOM, FF, NBR,
TR
Identifying Monetary Policy Shocks
•
•
•
•
Gali (1992)
CEE model little economic structure on the VAR beyond the monetary
policy reaction function.
identification.
VAR in four variables,
– the growth rate of real GNP (or GDP) (Y),
– the quarterly change in a short-term interest rate (DFF),
– the real interest rate, which equals the interest rate minus the quarterly inflation
rate in the consumer price index (P), and
– the growth rate of the real money supply (MONEY), which equals the log of the
nominal money supply (M1) minus the log of the price level
•
Gali imposes three long-run restrictions on the VAR:
– money supply shocks do not affect output,
– money demand shocks do not affect output,
– spending shocks do not affect output.
Three short-run restrictions:
– money supply shocks do not affect output contemporaneously,
– money demand shocks do not affect output contemporaneously,
– and the price level does not enter the money supply equation
contemporaneously.
Monetary Policy: Analytical
Revisions
• What Happens When Economists or
Policymakers Revise Conceptual
Variables?
– Orphanides (AER 2001, JME 2003a,b)
– Orphanides and Van Norden (2002)
Output Gap
• Defined with reference to an estimate of trend
output
• New Keynesian Models: depend on the output
level in the absence of price rigidities.
• Two possible mistakes: a) actual output b) trend
• Orphanides (2003) counterfactual simulations:
Fed severely overestimated trend output in the
70’s
– Output gap was underestimated
– Monpol was too loose
– Great inflation
Orphanides (2003)
• Counterfactual dynamic simulations
– Taylor Rule
it   t  0.5 yt  0.5( t   )  r
T
set
 r 2
it   t  0.5 yt  0.5( t  2)  2
T
*
– Revised Taylor Rule
• Change the output gap coefficient to 1!
*
Without noise
Orphanides (2003)
inflation accelerated in the late
1960s and 1970s because..
1. Fed deviated from the Taylor rule !
or
2. actually followed a strategy
indistinguishable from the Taylor rule!
One Possible Solution
• Alter the dataset CB reacts to
• Orphanides (2003) and Orphanides and
van Norden (2002) what matters is the
measurement of trend not the actual data
• Errors in measuring trend output are
persistent
– A look at ‘output gap first difference’ rule is
likely to be less sensitive than ‘output gap
levels’
Walsh (2004)
Walsh experiment (US Output gap 1959:1 2003:1)
• x0t=xt+qt
where x0t is estimated
gap, qt error
The trouble
• First difference seems to be better
alleviating measurement problem, but
• It is the level not the difference that enters
into the CB objective function
• There are New Keynesian models (Walsh
(2003) still compatible with this or
empirical evidence in favour of. (Mehra
2002 or Erceg and Levin (2003)
Measurement of Monetary
Aggregates
• One of the major puzzles in
macroeconomics: an apparent disconnect
between US monetary aggregates,
inflation and output
• Even the Fed Funds rate, while it can
account for US output variations, it has
little to say for the US inflation
Statistical Evidence
Granger Causality Tests
4
4
i 1
i 1
y      y    m  v
t
i t i
i
t i
t
VAR´s and Variance Decompositions
Stability Tests
B Friedman and Kuttner (1992) American Economic
Review, pp. 472-92.
and inflation….
What if monetary aggregates are measured badly?
Copyright © 2002
Pearson Education, Inc.
Slide 68
Aksoy&Piskorski (2006)
The relationship is stable for
output..
And for inflation..
Conclusion
• Nontrivial issue, data uncertainty is
important in both theoretical and empirical
applications