Upload
munin
View
26
Download
2
Tags:
Embed Size (px)
DESCRIPTION
Content Analysis of the IMF Article IV Staff Reports for Euro Area Countries. Lena Golubovskaja, NUIM April 14, 2011. IEA ANNUAL CONFERENCE. Outline. I ntroduction T heoretical Developments D ata and Methodology A nalysis of the Data C oncluding Remarks. Problem. The IMF’s Judgement. - PowerPoint PPT Presentation
Citation preview
Content Analysis of the IMF Article IV Staff Reports for Euro Area
Countries
Content Analysis of the IMF Article IV Staff Reports for Euro Area
Countries
Lena Golubovskaja, NUIM
April 14, 2011
IEA ANNUAL CONFERENCE
Outline
• Introduction
• Theoretical Developments
• Data and Methodology
• Analysis of the Data
• Concluding Remarks
• In the United States “core commercial and investment banks are in a sound financial position, and systemic risks appear low” (IMF, 2007:14)
• “Favorable global economic prospects, particularly strong momentum in the Euro area and in emerging markets led by China and India, continue to serve as a strong foundation for global financial stability…” (GFSR)
Problem
The IMF’s Judgement
EMU Fiscal Policies
Was the miracle a ‘mirage’?• Some northern countries (Germany,
Austria, the Netherlands) experienced competitiveness gains and accumulated huge external surpluses.
• Some southern countries (Spain, Portugal, Greece) accumulated huge external deficits under imbalanced high growth strategies driven by strong negative real interest rates.
Case Study: Ireland
• Large transitory tax receipts used to fund tax cuts and permanent spending commitments
• Budget surpluses were insufficiently large
• Excessive current spending growth
• Narrowed tax base
• Policy stance was inappropriately procyclical
Chart plots change in budget balance (x-axis) and in gross debt (y-axis), as %of GDP between 2007 and 2009. Source: Public Finances in EMU, EC 2009
-10.0
0.0
10.0
20.0
30.0
40.0
-15.0 -10.0 -5.0 0.0 5.0
IRE
LAT
ESP
UK
Deterioration in Fiscal Positions: EU Members, 2007-09
-10.0
0.0
10.0
20.0
30.0
40.0
-15.0 -10.0 -5.0 0.0 5.0
IRE
LAT
ESP
UK
Theoretical Developments
•Corporate earnings releases (Rogers at al., 2009)
• Accounting policy disclosures (Levine and Smith, 2006)
•Multiple sources of financial text (Kothari, Li, and Short, 2008)
•Asymmetric Effect of the Beige Book Language (Armesto et al., 2006)
Textual analysis using computer algorithms
Economic implications of the external policy
reports
The accuracy of information (Batchelor, 2001)
Pellechio and Cady (2005)
‘Anecdotical’ Evidence: Opinions of the IMF executive Board, Ireland
• “The continued impressive performance of the Irish economy, which is based on sound economic policies, provides useful lessons for other countries” (2004)
• “Impressive performance is due in significant measure to sound economic policies, including prudent fiscal policies… The housing market is likely headed for a soft landing, but there is a risk of a sharp decline.” (2005)
• “A number of directors, however, saw less merit in fiscal tightening at the current juncture, pointing to the need for further increases in spending to achieve social goals” (2006)
• “The directors commended Ireland’s continued impressive economic performance. this performance has been underpinned by outward-oriented policies, prudent fiscal policy, low taxes and labour market flexibility” (2007)
Content Analysis
1999 2000 2001 2002 2003 2004 2005 2006 2007 2009
Inappropriate fiscal stance
X X X X X X
Inflexibility of NWAs X X X
Declining competitiveness
X X X X X
House price overvaluation
X X X
Unwinding of construction boom
X X X X
Unbalanced growth X X
Vulnerability to external shocks
X X X
Vulnerability of banking system
X X X
Score 4 1 2 2 2 1 3 4 3 7
Concerns/Risk Factors Identified by IMF (IV), Ireland
Limitations*Author calculations
DICTION 5.0
• Dictionary-based program that counts types of words most frequently encountered in contemporary American public discourse and is designed to capture the linguistic style.
• Uses 10,000 search words assigned to 35 theoretically-based linguistic categories to characterize text on several dimensions.
• Advantages: pre-existing search rules, efficient analysis of large number of text, partial correction for the context.
• Disadvantage: assumes that higher frequency usages of a word or phrase mean the concept is more meaningful.
Content Analysis
DICTION 5.0: thematic categories and sub-categories
Content Analysis
DICTION 5.0: Hardship Dictionary
• Bias, Corrupting, Crime, Crises, Deficit, Depress, Deteriorate, Difficulties, Distress, Disturbance, Folly, Problem, Recession, Risk, Shock, Weakness, Worry, Worse…
Content Analysis
Data and Methodology
Data• 16 euro area countries• Staff Reports, IV (2005-2007)• Executive Board Assessments , PIN (2005-
2007)
Sample construction• Only textual part of the documents
Mean of 5.86 and standard deviation of 4.64 based on 122 runs of a variety of financial news stories.
Validation of WARNING measure
Year Score Z-score t-test p-value
2005 4.750 -0.240 2.6423 0.0093
2006 7.210 0.290 -3.2136 0.0017
2007 6.630 0.160 -1.8330 0.0693
2008 6.090 0.050 -0.5475 0.5850
2009 13.94 1.750 -19.2342 0.0000
Case of U.S., Staff Reports (IV)
WARNING: Euro AreaYear IV PIN t-test p-value
1999 4.480 3.669 0.9665 0.3448
2000 5.644 2.481 4.6893 0.0001
2001 5.352 3.178 2.5647 0.0167
2002 6.923 3.814 2.4581 0.0216
2003 6.111 3.430 2.7135 0.0117
2004 4.584 3.000 2.3867 0.0249
2005 5.404 3.662 2.4122 0.0235
2006 7.110 3.897 3.4015 0.0027
2007 6.309 2.975 3.6082 0.0013
2008 7.146 2.758 4.2968 0.0006
2009 9.423 4.395 6.1878 0.0000
Total 6.155 3.399 10.4324 0.0000
WARNING, Staff Reports
Average WARNING scores for the euro area: Staff Reports, 2005-2007
Insignificantly different results for Finland, France, Germany, Ireland, the Netherlands, and Slovakia
0.000
1.000
2.000
3.000
4.000
5.000
6.000
7.000
8.000
9.000
Austri
a
Belgium
Cypru
s
Finla
nd
Fran
ce
Germ
any
Greec
e
Irelan
dIta
ly
Luxe
mbo
urg
Malt
a
Nethe
rland
s
Portu
gal
Slovak
ia
Sloven
iaSpa
in
Sc
ore
Did WARNING score change over time for EMU? Staff Reports, 1999-2009
Year N Mean Z-score Std deviation t-test p-value
1999 9 4.480 -0.298 1.138
2000 14 5.644 -0.047 2.447 -1.3294 0.1980
2001 13 5.352 -0.110 2.459 0.3091 0.7598
2002 13 6.923 0.228 3.893 -1.2302 0.2306
2003 14 6.111 0.053 2.853 0.6215 0.5399
2004 13 4.584 -0.276 2.055 1.5846 0.1256
2005 14 5.404 -0.099 1.538 -1.1797 0.2492
2006 11 7.110 0.268 2.235 -2.2604 0.0336
2007 14 6.309 0.096 2.076 0.9261 0.3640
2008 9 7.146 0.276 2.809 -0.8224 0.4201
2009 10 9.423 0.767 1.877 -2.0982 0.0511
Total 134 6.155 2.661
WARNING, Staff Reports
WARNING for Ireland is not statistically different from the euro area averages only for the years 2002, 2003, and 2006
0
2
4
6
8
10
12
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
Sco
re WARNING, euro area
WARNING, Ireland
Summary
•Do not allow editing by member-country
•The language of surveillance needs to be unambiguous
•More weight on analysis, less on diplomacy: strengthen the role of staff/secretariat.
•Standard-based approach for staff interactions.