[PPT] - Early-warning indicators for debt sustainability Casper van Ewijk PowerPoint Presentation

SLIDE 1

Early-warning indicators for debt sustainability

Casper van Ewijk Jasper Lukkezen Hugo Rojas-Romagosa

SLIDE 2

Our main message in the words of Rudi Dornbusch

Think of someone who has made a great expertise of drunk driving, regularly drives drunk, tells you that he never has a problem, and one day there is a terrible, terrible accident. And he’ll say, "Well, it was the red light. It wasn’t my being

drunk. Normally that light is green."

CPB Netherlands Bureau for Economic Policy Analysis 2/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 3

How to assess debt sustainability?

A sustainable fiscal policy can be continued without losing control
ver the debt level
Towards stochastic analysis

◮ macro-volatility of interest and growth (economic uncertainty) ◮ response of fiscal policy to setbacks (policy maker)

Indicator captures upward risk of the debt level

◮ Expected debt increase which happens every 40 years ◮ In 2007, indicator identifies countries with sustainability issues ◮ Complements SGP (3%, 60%) and ageing (S1, S2) indicators

CPB Netherlands Bureau for Economic Policy Analysis 3/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 4

Stochastic analysis

Which government is more ’in control of its debt level’?

0% 20% 40% 60% 80% 100% 120% 140% 2007 2012 2017 2022 2027

Belgium

80% 100% 120% 140%

Portugal

0% 20% 40% 60% 80% 2007 2012 2017 2022 2027 2007 2012 2017 2022 2027

CPB Netherlands Bureau for Economic Policy Analysis 4/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 5

’at risk’ indicator captures upward risk

CPB Netherlands Bureau for Economic Policy Analysis 5/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 6

Rest of the presentation

1. What drives the debt level?
2. Theoretical debt sustainability: Modified Aaron condition
3. Stochastic simulations
4. The added value of the indicator

CPB Netherlands Bureau for Economic Policy Analysis 6/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 7

What drives the debt level?

Accounting equation for the debt level:

debtt+1 = 1 + interestt

1 + growtht × debtt − primary surplust.

Contributing channels
1. Growth
2. Interest
3. Surplus (fiscal response)

CPB Netherlands Bureau for Economic Policy Analysis 7/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 8

Autonomous debt reduction till 80s

Interest minus growth rate

.9 .95 1 1.05

Gamma parameter

1948 1951 1954 1957 1960 1963 1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 2011 year

The Netherlands

CPB Netherlands Bureau for Economic Policy Analysis 8/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 9

What drives the debt level?

Accounting equation for the debt level:

debtt+1 = 1 + interestt

1 + growtht × debtt − primary surplust.

Contributing channels
1. Growth
2. Interest
3. Surplus (fiscal response)

CPB Netherlands Bureau for Economic Policy Analysis 9/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 10

When Belgian debt increased, government budget responded

Debt ratio Primary surplus

.4 .6 .8 1 1.2 1.4 Debt / GDP 1955 1958 1961 1964 1967 1970 1973 1976 1979 1982 1985 1988 1991 1994 1997 2000 2003 2006 2009 year

Belgium

−.1 −.05 .05 .1 Primary Surplus / GDP 1955 1958 1961 1964 1967 1970 1973 1976 1979 1982 1985 1988 1991 1994 1997 2000 2003 2006 2009 year

Belgium CPB Netherlands Bureau for Economic Policy Analysis 10/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 11

When Portuguese debt increased, government budget did not respond

Debt ratio Primary surplus

.2 .4 .6 .8 1

Debt / GDP

1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 year

Portugal

−.1 −.05 .05

Primary Surplus / GDP

1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 year

Portugal

CPB Netherlands Bureau for Economic Policy Analysis 11/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 12

Theoretical debt sustainability: Modified Aaron condition

Modified Aaron condition: interest − growth − fiscal response < 0. with fiscal response the estimated responsiveness of surplus to debt. Then: If this condition is satisfied, debt converges to a steady state.

CPB Netherlands Bureau for Economic Policy Analysis 12/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 13

Data, simulation method & results

CPB Netherlands Bureau for Economic Policy Analysis 13/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 14

Data: long time series for fiscal response estimation

Main results: Post-WW2 data
Robustness: entire sample

Country Sample Observations USA 1792-2011 220 GBR 1691-2011 321 NLD 1816-2011* 188 BEL 1830-2011* 160 DEU 1970-2011 42 ITA 1862-2011 150 ESP 1850-2011* 159 PRT 1852-2011 160 ISL 1908-2011 103 * = War data missing

CPB Netherlands Bureau for Economic Policy Analysis 14/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 15

Modified Aaron condition satisfied for all countries

CPB Netherlands Bureau for Economic Policy Analysis 15/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 16

Historically debt was sustainable

High growth and low real interest contributes to sustainability
Fiscal response significant and positive for USA, GBR, NLD, BEL,

DEU and ITA not significant for ESP , PRT and ISL

Fiscal response robust when pre-WWII years are included

⇒ measures persistent institutional characteristic

CPB Netherlands Bureau for Economic Policy Analysis 16/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 17

Since mid 80s: fiscal response required

growth > interest prior to 1987 and interest > growth afterwards

⇒ fiscal response required for sustainability

How bad is it?

⇒ Simulation needed

CPB Netherlands Bureau for Economic Policy Analysis 17/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 18

Volatility much higher in Iceland

USA ISL

−.1 .1 .2

real growth

1950 1975 2000

years

−.1 .1 .2

real growth

1950 1975 2000

years

−.3 −.1 .1

real interest

1950 1975 2000

years

−.3 −.1 .1

real interest

1950 1975 2000

years CPB Netherlands Bureau for Economic Policy Analysis 18/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 19

Stochastic simulation

Simulate debt going forward:

1. Estimate the fiscal response
2. Simulate volatility in interest and growth rates (Budina and van

Wijnbergen, 2008) using a VAR

3. Simulate debt at time t + 1 from time t debt, simulated volatility in

interest and growth and estimated fiscal response

CPB Netherlands Bureau for Economic Policy Analysis 19/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 20

Fiscal response reduces debt levels and volatility

80% 100% 120% 140%

Germany, response=0

0% 20% 40% 60% 2011 2016 2021 80% 100% 120% 140%

Germany, response=2.6%

0% 20% 40% 60% 2011 2016 2021

CPB Netherlands Bureau for Economic Policy Analysis 20/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 21

Volatility in interest and growth increase debt volatility

80% 100% 120% 140%

Germany, response=2.6%

0% 20% 40% 60% 2011 2016 2021 80% 100% 120% 140%

Spain, response=0

0% 20% 40% 60% 2011 2016 2021

CPB Netherlands Bureau for Economic Policy Analysis 21/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 22

Simulation outcomes

Larger fiscal response reduces debt levels
Larger fiscal response and smaller interest and growth rate

volatility reduce debt volatility

Define ’at risk’ indicator: debt level that is higher then 97.5% of

the debt levels minus median debt level after 10 years. Remaining 2.5% ≈> once every 40 years

CPB Netherlands Bureau for Economic Policy Analysis 22/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 23

’at risk’ indicator

CPB Netherlands Bureau for Economic Policy Analysis 23/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 24

2011 indicator

CPB Netherlands Bureau for Economic Policy Analysis 24/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 25

Early-warning indicator

’07 indicator value is highly correlated with ’09-’12 sovereign

spreads.

’07 sovereign spreads are not correlated with ’09-’12 sovereign

spreads

CPB Netherlands Bureau for Economic Policy Analysis 25/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 26

ITAESP PRT ISL 200 300 400 500 600 an '09-'12 CDS rate USA GBR NLD BEL DEU 100 200 5 10 15 20 25 30 35 Mea 'at risk'-indicator in '07

SLIDE 27

Discussion

CPB Netherlands Bureau for Economic Policy Analysis 27/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 28

How to use the indicators?

Assess whether fiscal response is sufficient to sustain medium

term debt levels when other instruments are absent

NOT: Ability to refinance or probability of liquidity crisis

Complements current set of indicators:

Debt-level (SGP norms)
Structural balances
Ageing study sustainability indicators

CPB Netherlands Bureau for Economic Policy Analysis 28/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 29

Take aways

Medium term debt sustainability depends on macro-volatility

⇒ Stochastic simulation required

And country specific response of fiscal policy
Our framework first step towards full stochastic analysis
The ’at risk’- indicator distinguishes countries with sustainability

issues (ITA, ESP , PRT) from countries without (USA, GBR, NLD, BEL)

CPB Netherlands Bureau for Economic Policy Analysis 29/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 30

Thank you for your attention!

CPB Netherlands Bureau for Economic Policy Analysis 30/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 31

Bibliography I

Bohn, H. (2007). “Are Stationary and Cointegration Restrictions Really Necessary for the IntertemporalBudget Constraing?”, Journal of Monetary Economics 54(7): 1837–1847. Budina, N. and van Wijnbergen, S. (2008). “Quantitative Approaches to Fiscal Sustainability Analysis: A Case Study of Turkey since the Crisis of 2001”, World Bank Economic Review 23(1): 119–140.

CPB Netherlands Bureau for Economic Policy Analysis 31/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 32

SLIDE 33

Estimating the fiscal response

Estimation:

primary surplust = α + fiscal response × debtt + β othert + εt.

’Other’ corrects for:

◮ Business cycle ◮ Temporary government spending

Fiscal response:

◮ Measures long-term response of fiscal policy to government debt ◮ Indicates whether governments reduce their debt over time

CPB Netherlands Bureau for Economic Policy Analysis 33/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 34

Stationarity conditions

From Bohn (2007) we know:

δ < 1

Debt stationary, steady state −γα/(1 − δ)

1 < δ < ρ

Debt explosive but consistent with the IBC

1 < δ and ρ < δ

Debt explosive For δ = 1+r

1+y(1 − ρ)

CPB Netherlands Bureau for Economic Policy Analysis 34/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 35

Fiscal response

Estimation:

st = α + ρdt + βZt + εt.

With Zt:

◮ Business cycle ◮ Temporary government spending

ρ is the fiscal reaction

Measure of long-term response of fiscal policy to government debt. It indicates whether governments are willing to reduce their debt level

ver time in the absence of outside pressure to do so.

CPB Netherlands Bureau for Economic Policy Analysis 35/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 36

.1 .2 .3 .4 .5

Military expenditure / GDP

1691 1711 1731 1751 1771 1791 1811 1831 1851 1871 1891 1911 1931 1951 1971 1991 2011 year

United Kingdom

SLIDE 37

50% 52% 54% 56% 58% 60% 2009 2010 2011 2012 2013 2014

Debt/GDP

United States

rho=0 rho>0 rir shock

60% 65% 70% 75% 80% 85% 90% 2009 2010 2011 2012 2013 2014

Debt/GDP

United Kingdom

rho=0 rho>0 rir shock

SLIDE 38

Empirical procedure

1. Estimate the fiscal response:

st = α + ρdt + βZt + εt.

2. Estimate a VAR for interest and growth rates (Budina and van

Wijnbergen, 2008):

rt yt

= α0 +

∞

∑

j=1

Aj rt−j yt−j

+ ηt,

var (ηt) = Σ.

3. Simulate debt at time t + 1 from time t data:

dt+1 = 1 + rt 1 + yt (1 − ρ) dt − γα,

using the VAR shocks and the fiscal response coefficients.

CPB Netherlands Bureau for Economic Policy Analysis 38/30 Lunchtalk at Bruegel Early-warning indicators | 8 October 2013

SLIDE 39

.2 .4 .6 .8 Debt / GDP 1948 1951 1954 1957 1960 1963 1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 year

United States

.5 1 1.5 2 2.5 Debt / GDP 1946 1949 1952 1955 1958 1961 1964 1967 1970 1973 1976 1979 1982 1985 1988 1991 1994 1997 2000 2003 2006 2009 year

United Kingdom

.5 1 1.5 2 Debt / GDP 1948 1951 1954 1957 1960 1963 1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 2011 year

The Netherlands

.4 .6 .8 1 1.2 1.4 Debt / GDP 1955 1958 1961 1964 1967 1970 1973 1976 1979 1982 1985 1988 1991 1994 1997 2000 2003 2006 2009 year

Belgium

.2 .4 .6 .8 Debt / GDP 1970 1973 1976 1979 1982 1985 1988 1991 1994 1997 2000 2003 2006 2009 2012 year

Germany

.2 .4 .6 .8 1 1.2 Debt / GDP 1948 1951 1954 1957 1960 1963 1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 year

Italy

.2 .4 .6 .8 Debt / GDP 1945 1948 1951 1954 1957 1960 1963 1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 2011 year

Spain

.2 .4 .6 .8 Debt / GDP 1945 1948 1951 1954 1957 1960 1963 1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 2011 year

Portugal

.2 .4 .6 .8 1 Debt / GDP 1946 1949 1952 1955 1958 1961 1964 1967 1970 1973 1976 1979 1982 1985 1988 1991 1994 1997 2000 2003 2006 2009 year

Iceland

SLIDE 40

.92 .94 .96 .98 1 1.02 Gamma parameter 1948 1951 1954 1957 1960 1963 1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 year

United States

.96 .98 1 1.02 1.04 Gamma parameter 1946 1949 1952 1955 1958 1961 1964 1967 1970 1973 1976 1979 1982 1985 1988 1991 1994 1997 2000 2003 2006 2009 year

United Kingdom

.9 .95 1 1.05 Gamma parameter 1948 1951 1954 1957 1960 1963 1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 2011 year

The Netherlands

.96 .98 1 1.02 1.04 Gamma parameter 1955 1958 1961 1964 1967 1970 1973 1976 1979 1982 1985 1988 1991 1994 1997 2000 2003 2006 2009 year

Belgium

.98 .99 1 1.01 1.02 1.03 Gamma parameter 1970 1973 1976 1979 1982 1985 1988 1991 1994 1997 2000 2003 2006 2009 2012 year

Germany

.7 .8 .9 1 Gamma parameter 1948 1951 1954 1957 1960 1963 1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 year

Italy

.85 .9 .95 1 1.05 1.1 Gamma parameter 1945 1948 1951 1954 1957 1960 1963 1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 2011 year

Spain

.8 .85 .9 .95 1 Gamma parameter 1945 1948 1951 1954 1957 1960 1963 1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 2011 year

Portugal

.7 .8 .9 1 Gamma parameter 1946 1949 1952 1955 1958 1961 1964 1967 1970 1973 1976 1979 1982 1985 1988 1991 1994 1997 2000 2003 2006 2009 year

Iceland

SLIDE 41

Countries with ρ significant

80% 100% 120% 140% United States, ρ=7.8% 0% 20% 40% 60% 2011 2016 2021 2026 2031 80% 100% 120% 140% United Kingdom, ρ=4.5% 0% 20% 40% 60% 2011 2016 2021 2026 2031 80% 100% 120% 140% Netherlands, ρ=7.7% 0% 20% 40% 60% 2011 2016 2021 2026 2031 80% 100% 120% 140% Belgium, ρ=3.8% 0% 20% 40% 60% 2011 2016 2021 2026 2031 80% 100% 120% 140% Germany, ρ=2.6% 0% 20% 40% 60% 2011 2016 2021 2026 2031 80% 100% 120% 140% Italy, ρ=7.1% 0% 20% 40% 60% 2011 2016 2021 2026 2031

Countries with ρ not significant, ρ = 0.07 assumed

80% 100% 120% 140% Spain, ρ=7.0% 0% 20% 40% 60% 2011 2016 2021 2026 2031 80% 100% 120% 140% Portugal, ρ=7.0% 0% 20% 40% 60% 2011 2016 2021 2026 2031 80% 100% 120% 140% Iceland, ρ=7.0% 0% 20% 40% 60% 2011 2016 2021 2026 2031

SLIDE 42

Countries with ρ significant, ρ = 0 assumed

80% 100% 120% 140% United States, ρ=0 0% 20% 40% 60% 2011 2016 2021 2026 2031 80% 100% 120% 140% United Kingdom, ρ=0 0% 20% 40% 60% 2011 2016 2021 2026 2031 80% 100% 120% 140% Netherlands, ρ=0 0% 20% 40% 60% 2011 2016 2021 2026 2031 80% 100% 120% 140% Belgium, ρ=0 0% 20% 40% 60% 2011 2016 2021 2026 2031 80% 100% 120% 140% Germany, ρ=0 0% 20% 40% 60% 2011 2016 2021 2026 2031 80% 100% 120% 140% Italy, ρ=0 0% 20% 40% 60% 2011 2016 2021 2026 2031

Countries with ρ not significant, ρ = 0

80% 100% 120% 140% Spain, ρ=0 0% 20% 40% 60% 2011 2016 2021 2026 2031 80% 100% 120% 140% Portugal, ρ=0 0% 20% 40% 60% 2011 2016 2021 2026 2031 80% 100% 120% 140% Iceland, ρ=0 0% 20% 40% 60% 2011 2016 2021 2026 2031