Transcript Numerical Fiscal Rules and the Response of

The Impact of Expenditure Rules on Budgetary
Discipline over the Cycle
Sebastian Hauptmeier
Fédéric Holm-Hadualla
Philipp Rother
(European Central Bank, Fiscal Policies Division)
44th Annual Conference of the CEA
Quebec City, 28 May 2010
Aim of this paper
Test the hypothesis that …
…“enforced national expenditure rules (…) help to counteract
forces leading to pro-cyclical fiscal policy in good times and thus
prevent the need to retrench in bad times.”
(European Commission, Quarterly Report on the Euro Area, 2004)
2
Theoretical Motivation
A model of spending indiscipline
Starting point: common cool problem in budgetary decisionmaking (von Hagen and Harden, 1995)
Assumptions:
• budgetary process is fragmented
• decision makers cater to diverse constituencies
Implication: imperfect overlap between
• subset of electorate which benefits from a particular spending
programme and
• subset bearing the associated cost
 Tendency for inefficiently high levels of spending and procyclical expenditure policies (Tornell and Lane, 1999)
3
Theoretical Motivation
The role of numerical fiscal rules
Two interpretations of the role of numerical fiscal rules
1. In a narrow sense: strictly binding constraints on fiscal policy
(see e.g. Kopitz and Symanski, 1998)
2. In a broader sense: benchmark against which imperfectly
informed electorate assesses governments’ fiscal performance
(see e.g. Schuknecht, 2004)
 Numerical rules may enhance governments’ fiscal discipline
4
Theoretical Motivation
The rationale for separately analyzing individual expenditure items
Related empirical literature relies on broad fiscal aggregates as
dependent variables (e.g. total or primary expenditure)
Hypothesis of a pro-cyclical spending bias rests on the assumption
that governments react spontaneously to exogenous shocks
In practice overall expenditure includes items characterised by a
lack of budgetary flexibility (in the short-run)
 Non-discretionary expenditure items may partly conceal the
pro-cyclical spending bias
5
Research Questions
1. Do governments deviate from expenditure plans in response
to surprising changes in the cyclical position of the economy
2. If yes, do expenditure rules shape this reaction?
3. Do the answers to 1. and 2. differ depending on which
expenditure items are analysed?
6
Empirical Analysis
Dependent variables
Main idea (see Strauch et al. 2004)
• interpret expenditure forecasts reported in Stability and
Convergence Programmes from t-1 as target
• compute difference to actual outcome according to latest
revision (based on Ameco data)
dev  g  g
k
k
k
Implementation
i ,t
i ,t
i ,t
• where k denotes
– Total or primary spending
– “Discretionary” spending
– “Non-Discretionary” spending (all as ratio of GDP)
• unbalanced panel of EU15 (EU25) countries from 2002 – 2008
(2004-2008); data from 145 SCP
7
Empirical Analysis
Discriminating between discretionary and non-discretionary spending
Ideal setup
• subsume all expenditure items under these categories
• compute one dependent variable for each category
Problem: some SCP categories comprise both types of spending
Feasible setup
• only choose variables not affected by this problem i.e.:
– gross fixed capital formation and subsidies for discretionary
– interest payments for non-discretionary expenditure
8
Empirical Analysis
Main explanatory variables (1): output gap surprise
Main idea: (Poterba, 1994) capture the deviations between
• governments’ expectations on cyclical developments when
setting their targets and
• the actual outcome according to latest revision
OG  h  h
Implementation
i ,t
i ,t
i ,t
• where
 h denotes revised output gap (from Ameco)
i ,t

hi ,t denotes the European Commission’s autumn forecast
(from period t-1) for the output gap in period t
(both variables are in percent of trend GDP)
9
Empirical Analysis
Main explanatory variables (2): expenditure rules index
Main idea: translate a broad set of institutional provisions into a
country-specific cardinal ranking
Implementation: expenditure rule index (see Debrun et al. 2008)
which captures
• the share of general government spending covered by
expenditure rules and
• qualitative features of rules relevant for their effectiveness
(statutory basis, enforcement mechanism, media visibility)
10
Econometric Analysis
Descriptive Statistics
Dependent variables
Deviation Total Expenditure
Deviation Primary Expenditure
Deviation Discretionary Expenditure
Deviation Interest Expenditure
Explanatory variables
Output Gap Surprise
Expenditure Rules Index
Debt Ratio
Expenditure Ratio
Revenue Ratio
Budget balance ratio
GDP deflator
Obs.
Mean
Std. Dev.
Min.
Max.
145
145
139
145
1.07
1.12
0.09
-0.54
1.75
1.73
0.68
0.28
-7.03
-6.80
-2.22
-0.69
6.39
6.28
-2.75
1.32
145
145
145
145
145
145
145
0.71
0
50.33
45.04
43.89
-1.14
3.63
1.25
1
26.62
5.90
6.29
2.73
3.28
-2.88
-0.79
3.49
33.34
32.04
-9.32
-0.72
5.43
2.27
106.89
56.95
57.69
5.33
20.30
Note: all variables except for GDP deflator and expenditure rules index are expressed in % of GDP.
GDP deflator is measured in terms of percentage change on preceding year.
11
Econometric Analysis
Specification
dev  ci  dt   OGi ,t   (OGi ,t  ERi )   X i ,t  u
k
i ,t
k
i ,t
where
• OGi ,t denotes output gap surprises
• ERi measures the strength of expenditure rules
• X i ,t matrix of controls to capture
 fiscal position
 position in the electoral cycle
• ci and dt are country and time fixed effects, respectively, and
• uik,t denotes the error term
12
Econometric Analysis
Instrumenting the output gap surprise
Problem: potential simultaneity of output gap surprise and
government spending
Identification strategy: use “structural” part of output gap surprise
as an instrument
• deviations between forecast and may arise due to actual
and/or trend GDP
• the latter should – by definition – not be affected by short-run
economic fluctuations from government spending
 Instrumentation of output gap surprise with forecast error in
trend GDP
Additional instrument: avg. output gap in other sample countries
13
Econometric Analysis
Baseline Results
Output Gap Surprise (OGS)
(OGS)*(Expenditure Rules Index)
Government Debt (first lag)
Total Government Spending (first lag)
Observations
Sargan/Hansen test of overidentifying
restrictions (p-value)
Difference in Sargan/Hansen test for
endogenous regressor (p-value)
Kleinbergen-Paap rk Wald F-Statistic
(Critical value Stock-Yogo weak
identification test for 5% maximal IV
relative bias: 11.4
and 10% maximal distortion of Wald test:
16.9)
(1)
(2)
(3)
(4)
Deviation
Total
Expenditure
Deviation
Primary
Expenditure
Deviation
Discretionary
Expenditure
Deviation
Interest
Expenditure
0.44*
(0.24)
-0.28*
(0.16)
0.07**
(0.03)
-0.05
(0.11)
145
0.47**
(0.24)
-0.31**
(0.16)
0.09***
(0.03)
-0.03
(0.11)
145
0.34***
(0.08)
-0.11**
(0.04)
0.02*
(0.01)
0.01
(0.04)
139
-0.03
(0.04)
0.02
(0.02)
-0.01***
(0.00)
-0.01
(0.02)
145
0.43
0.69
0.94
0.08
0.13
0.10
0.45
0.56
32.93
32.93
34.56
32.93
14
Econometric Analysis
Baseline Results
Output Gap Surprise (OGS)
(OGS)*(Expenditure Rules Index)
Government Debt (first lag)
Primary Government Spending (first lag)
Parliamentary Election
(1)
(2)
(3)
(4)
Deviation
Primary
Expenditure
Deviation
Discretionary
Expenditure
Deviation
Primary
Expenditure
Deviation
Discretionary
Expenditure
0.46**
(0.24)
-0.28*
(0.17)
0.10***
(0.03)
-0.06
(0.11)
0.48*
(0.26)
0.35***
(0.08)
-0.10**
(-0.04)
0.02*
(0.01)
0.01
(0.04)
0.07
(0.08)
0.08***
(2.72)
-0.09
(-0.81)
0.02*
(1.85)
0.01
(0.20)
0.59**
(2.07)
0.08
(0.11)
-0.17
(-0.68)
-0.58
0.37***
(3.46)
0.17
(0.70)
-0.14**
(-2.21)
-0.01
(-1.29)
(-0.08)
145
139
OGS if Output Gap > 0
OGS if Output Gap < 0
(OGS)*(Expenditure Rules Index) if Output Gap > 0
(OGS)*(Expenditure Rules Index) if Output Gap < 0
Observations
145
139
15
Conclusion
1.
The results confirm
•
the need for institutional restrictions to expenditure
policy
•
as well as their effectiveness
2.
Analysis points to a trade-off with respect to the institutional
design of these rules since
•
including non-discretionary spending items adds “noise”
•
excluding non-discretionary spending items
–
reinforces politicians’ information advantage
–
introduces a “status-quo bias”
16
Econometric Analysis
Robustness Checks
Output Gap Surprise (OGS)
(OGS)*(Expenditure Rules Index)
Government Debt (first lag)
Total Government Revenue (first lag)
(1)
Deviation
Total
Expenditure
0.49**
(0.23)
-0.25
(0.16)
0.07**
(0.03)
0.05
(0.10)
(2)
Deviation
Primary
Expenditure
0.51**
(0.24)
-0.30*
(-0.16)
0.08***
(0.03)
0.03
(0.10)
(3)
Deviation
Discretionary
Expenditure
0.34***
(0.07)
-0.11***
(0.04)
0.02*
(0.01)
-0.01
-0.04
General Government budget balance (first lag)
Observations
Sargan/Hansen test of overidentifying restrictions (pvalue)
Difference in Sargan/Hansen test for suspect
instruments (p-value)
Difference in Sargan/Hansen test for endogenous
regressor (p-value)
Kleinbergen-Paap rk Wald F-Statistic
(Critical value Stock-Yogo weak identification test for
5% maximal IV relative bias: 11.4
and 10% maximal distortion of Wald test: 16.9)
(4)
Deviation
Total
Expenditure
0.46*
(0.24)
-0.28*
(0.16)
0.08**
(0.03)
(5)
Deviation
Primary
Expenditure
0.48**
(0.25)
-0.32*
(0.17)
0.09***
(0.03)
(6)
Deviation
Discretionary
Expenditure
0.34***
(0.08)
-0.11**
(0.04)
0.02*
(0.01)
0.06
(0.10)
145
-0.01
(0.04)
139
145
145
139
0.08
(0.11)
145
0.36
0.62
0.93
0.38
0.64
0.94
0.36
0.61
0.76
0.30
0.55
0.78
0.07
0.06
0.54
0.07
0.06
0.47
29.6
29.6
29.7
32.5
32.5
33.7
Note: All estimates are obtained from two-stage least squares estimation. Excluded instruments for the output gap surprise and its interaction with the expenditure rules index are
the forecast error in trend GDP, the average output gap in all other ountries in the sample and the corresponding interaction terms. Robust standard errors in parentheses. ***
significant at 1% level, ** significant at 5% level, * significant at 10% level.
17
Econometric Analysis
Robustness Checks
Output Gap Surprise (OGS)
(OGS)*(Expenditure Rules Index)
Government Debt (first lag)
Primary Government Spending (first lag)
(1)
Deviation
Primary
Expenditure
0.46*
(0.25)
-0.33**
(0.16)
0.09***
(0.03)
-0.07
(0.11)
Discretionary Government Spending (first lag)
(2)
Deviation
Discretionary
Expenditure
0.34***
(0.08)
-0.11***
(0.04)
0.02*
(0.01)
(4)
Deviation
Primary
Expenditure
0.46*
(1.91)
-0.31*
(0.17)
0.08***
(0.03)
(5)
Deviation
Discretionary
Expenditure
0.35***
(0.07)
-0.11***
(0.04)
0.02*
(1.74)
-0.04
(0.11)
-0.01
(0.08)
0.07
(0.16)
145
-0.03
(0.11)
-0.01
(0.09)
0.00
(0.09)
145
0.04
(0.04)
-0.07
(0.13)
Total Government Spending (first lag)
GDP deflator
Forecast Error in nominal GDP Growth
Observations
(3)
Deviation
Total
Expenditure
0.43*
(0.24)
-0.28*
(0.17)
0.07**
(2.29)
145
139
-0.06
(0.03)
0.14*
(0.08)
139
18