Cross-Country Heterogeneity in Intertemporal Substitution

Transcription

Motivation
Methodology
Results
Summary
Cross-Country Heterogeneity in Intertemporal
Substitution
Tomas Havranek
Roman Horvath
Marek Rusnak
Zuzana Irsova
Charles University, Institute of Economic Studies
Czech National Bank, Research Department
Joint Research Center for Dynamic Models in Economics
Workshop, 2 June 2014
Tomas Havranek (IES, CNB)
Elasticity of Intertemporal Substitution
DYME, 2 June 2014
1 / 23
Motivation
Methodology
Results
Summary
Elasticity of Intertemporal Substitution in Consumption
The EIS reflects households’ willingness to substitute
consumption between time periods in response to changes in
the expected real interest rate.
Important in models involving intertemporal choice:
• monetary policy,
• fiscal policy,
• portfolio choice,
• computing the social cost of carbon emissions, and more.
DYME, 2 June 2014
2 / 23
Motivation
Methodology
Results
Summary
The Elasticity Matters.
0.4
Change in investment (%)
0.2
0
-0.2
-0.4
EIS = 0.1
EIS = 0.3
EIS = 0.5
EIS = 1
EIS = 1.5
-0.6
-0.8
-1
0
2
4
6
8
10
12
14
16
Quarters after a one-percentage-point increase in the policy rate
18
DYME, 2 June 2014
3 / 23
Motivation
Methodology
Results
Summary
Calibrations Vary.
Study
EIS
Outlet
House & Shapiro (2006)
Piazzesi et al. (2007)
Chari et al. (2003)
Trabandt & Uhlig (2011)
Jin (2012)
Rudebusch & Swanson (2012)
Smets & Wouters (2007)
Bansal & Yaron (2004)
Ai (2010)
Barro (2009)
Colacito & Croce (2011)
0.2
0.2
0.2
0.5
0.5
0.5
2/3
1.5
2
2
2
American Economic Review
Journal of Financial Economics
Review of Economic Studies
Journal of Monetary Economics
American Economic Journal: Macro
Journal of Finance
Journal of Finance
Journal of Political Economy
DYME, 2 June 2014
4 / 23
Motivation
Methodology
Results
Summary
Different Justifications
Study
EIS
Explanation
House & Shapiro (2006)
0.2
Trabandt & Uhlig (2011)
0.5
Smets & Wouters (2007)
2/3
Barro (2009)
2
p. 1837: “Our calibration is roughly the
average estimate in Hall (1988), Campbell and Mankiw (1989), and Barsky et al.
(1997).”
p. 311: “For the intertemporal elasticity
of substitution, a general consensus is followed for it to be close to 0.5.”
p. 593: “These are all quite standard calibrations.”
p. 252: “Because of the shortcomings of
the macroeconomic estimates, it is worthwhile to consider microeconomic evidence.
The Gruber (2006) analysis is particularly
attractive (. . . ).”
DYME, 2 June 2014
5 / 23
Motivation
Methodology
Results
Summary
Call for a Meta-Analysis
Empirical studies produce very different estimates of the EIS.
Browning & Lusardi (1996, p. 1833):
• “It is frustrating in the extreme that we have very little idea
of what gives rise to the different findings.”
• “We still await a study which traces all of the sources of
differences in conclusions to sample period; sample
selection; functional form; variable definition; demographic
controls; econometric technique; stochastic specification;
instrument definition; etc.”
DYME, 2 June 2014
6 / 23
Motivation
Methodology
Results
Summary
Uses of Meta-Analysis
Havranek, T. & Z. Irsova (2011): Estimating Vertical
Spillovers from FDI: Why Results Vary and What the True
Effect Is.
Journal of International Economics 85(2): pp. 234–44.
Rusnak, M., T. Havranek, & R. Horvath (2013): How to
Solve the Price Puzzle? A Meta-Analysis.
Journal of Money, Credit and Banking 45(1): pp. 37–70.
Havranek, T., Z. Irsova, & K. Janda (2012): Demand for
Gasoline is More Price-Inelastic than Commonly Thought.
Energy Economics 34(1): pp. 201–207.
DYME, 2 June 2014
7 / 23
Motivation
Methodology
Results
Summary
Results
Main Findings
1
The elasticity of intertemporal substitution varies a lot
across countries.
2
Households in rich countries and countries with high stock
market participation substitute more intertemporally.
3
Some method choices have systematic effects on the
estimated elasticity (data frequency, inclusion of durables).
Project Website
www.meta-analysis.cz/substitution
DYME, 2 June 2014
8 / 23
Motivation
Methodology
Results
Summary
Estimating the Elasticity
Researchers usually follow Hall (1988) and use the
log-linearized consumption Euler equation:
∆ct+1 = αi + EIS · ri,t+1 + i,t+1 .
• ∆ct+1 is consumption growth at time t + 1
• ri,t+1 is the real return on asset i at time t + 1
Instruments for ri,t+1 typically include the values of asset
returns and consumption growth known at time t.
DYME, 2 June 2014
9 / 23
Motivation
Methodology
Results
Summary
Data Collection
• We find 169 published studies that report estimates of the
EIS.
• These studies provide in total 2735 estimates for 104
countries.
• We collect all estimates, their standard errors, 30 aspects
of methodology, and characteristics of the countries.
• The average is 0.5 (0.8 for micro estimates; 0.9 for studies
published in top journals).
DYME, 2 June 2014
10 / 23
Motivation
Methodology
Results
Summary
The Elasticity Varies Across Countries.
EIS > 0.7
EIS ∈ (0.5, 0.7]
EIS ∈ [0.3, 0.5]
EIS ∈ [0.1, 0.3)
EIS < 0.1
no data
DYME, 2 June 2014
11 / 23
Motivation
Methodology
Results
Summary
What Explains the Differences?
Country-level variables
Stock market participation: Euler equation valid for asset
holders.
GDP per capita: necessities hard to substitute across time
periods.
Credit availability: financial constraints hamper intertemporal
substitution.
Real interest: the elasticity does not have to be constant.
Rule of law: institutions can affect financial decisions.
DYME, 2 June 2014
12 / 23
Motivation
Methodology
Results
Summary
The Elasticity Varies Across Methods.
Bosca et al. (2006)
Campbell (1999)
Campbell (2003)
Campbell and Mankiw (1991)
Chyi and Huang (1997)
Fuse (2004)
Hamori (1996)
Ho (2004)
Ito and Noda (2012)
Jimenez−Martin and deFrutos (2009)
Kim and Ryou (2012)
Koedijk and Smant (1994)
Nieh and Ho (2006)
Noda and Sugiyama (2010)
Ogaki et al. (1996)
Okubo (2011)
Osano and Inoue (1991)
Pagano (2004)
Rodriguez et al. (2002)
Sakuragawa and Hosono (2010)
Sarantis and Stewart (2003)
Yogo (2004)
−5
0
5
estimate of the EIS
10
DYME, 2 June 2014
13 / 23
Motivation
Methodology
Results
Summary
Control Variables (1)
Utility
Epstein-Zin
Habits
Nonsep.
durables
Nonsep. public
Nonsep.
ables
trad-
Data
No. of households
No. of years
Average year
Micro data
Annual data
Monthly data
=1 if the estimation differentiates between the EIS and the
coefficient of relative risk aversion.
=1 if habits in consumption are assumed.
=1 if the model allows for nonseparability between durables
and nondurables.
=1 if the model allows for nonseparability between private and
public consumption.
=1 if the model allows for nonseparability between tradables
and nontradables.
The logarithm of the number of cross-sectional units used in
the estimation (households, cohorts, countries).
The logarithm of the number of years of the data period used
in the estimation.
The logarithm of the average year of the data period.
=1 if the coefficient comes from a micro-level estimation.
=1 if the data frequency is annual.
=1 if the data frequency is monthly.
DYME, 2 June 2014
14 / 23
Motivation
Methodology
Results
Summary
Design
Quasipanel
Inverse estimation
Asset holders
First lag instrument
No year dummies
Income
Taste shifters
Variable definition
Total consumption
Food
Stock return
Capital return
=1 if quasipanel (synthetic cohort) data are used.
=1 if the rate of return is the dependent variable in the estimation.
=1 if the estimate is related to the rich or asset holders.
=1 if the first lags of variables are included among instruments.
=1 if year dummies are omitted in micro studies using the
Panel Study of Income Dynamics.
=1 if income is included in the specification.
The logarithm of the number of controls for taste shifters.
=1 if total consumption is used in the estimation.
=1 if food is used as a proxy for nondurables.
=1 if the rate of return is measured as stock return.
=1 if the rate of return is measured as the return on capital.
DYME, 2 June 2014
15 / 23
Motivation
Methodology
Results
Summary
Estimation
Exact Euler
ML
TSLS
OLS
Publication
SE
Publication year
Citations
Top journal
Impact
=1 if the exact Euler equation is estimated.
=1 if maximum likelihood methods are used for estimation.
=1 if two-stage least squares are used for estimation.
=1 if ordinary least squares are used for estimation.
The reported standard error of the estimate of the EIS.
The logarithm of the year of publication of the study.
The logarithm of the number of per-year citations of the study
in Google Scholar.
=1 if the study was published in one of the top five journals in
economics.
The recursive RePEc impact factor of the outlet.
DYME, 2 June 2014
16 / 23
Motivation
Methodology
Results
Summary
Bayesian Model Averaging
Model Inclusion Based on Best 5000 Models
Stock market partic.
GDP per capita
Credit availability
Real interest
Rule of law
Inverse estimation
Top journal
No. of years
Total consumption
Stock return
OLS
Capital return
Citations
Asset holders
Nonsep. durables
Monthly data
Exact Euler
Quasipanel
Epstein-Zin
ML
Food
TSLS
First lag instrument
No. of households
No year dummies
Habits
Impact
Nonsep. tradables
Micro data
Income
Annual data
Publication year
Taste shifters
Average year
Nonsep. public
0
0.05
0.1 0.15 0.19 0.24
0.29
0.34 0.38
0.43 0.48 0.53 0.57 0.62 0.66 0.7 0.75 0.79 0.84 0.88 0.93
Cumulative Model Probabilities
DYME, 2 June 2014
17 / 23
Motivation
Methodology
Results
Summary
Posterior Coefficient Distributions (1)
(a) GDP per capita
(b) Credit availability
Marginal Density: Credit_availability (PIP 100 %)
7
Marginal Density: GDP_per_capita (PIP 100 %)
6
Cond. EV
2x Cond. SD
Median
Density
3
3
0
0
1
1
2
2
Density
4
5
4
5
Cond. EV
2x Cond. SD
Median
−0.2
−0.1
0.0
0.1
0.2
0.3
0.4
−0.3
−0.2
Coefficient
−0.1
0.0
0.1
Coefficient
DYME, 2 June 2014
18 / 23
Motivation
Methodology
Results
Summary
Posterior Coefficient Distributions (2)
(c) Real interest
(d) Rule of law
Marginal Density: Real_interest (PIP 100 %)
Marginal Density: Rule_of_law (PIP 100 %)
4
Cond. EV
2x Cond. SD
Median
2
Density
30
0
0
10
1
20
Density
3
40
50
Cond. EV
2x Cond. SD
Median
−0.03
−0.02
−0.01
0.00
0.01
0.02
−0.4
−0.2
Coefficient
0.0
0.2
Coefficient
DYME, 2 June 2014
19 / 23
Motivation
Methodology
Results
Summary
Stock Market Participation
Marginal Density: Stock_market_partic. (PIP 100 %)
Density
0.0
0.1
0.2
0.3
0.4
0.5
0.6
Cond. EV
2x Cond. SD
Median
0
1
2
3
4
5
Coefficient
DYME, 2 June 2014
20 / 23
Motivation
Methodology
Results
Summary
Economic Significance
Stock market participation and GDP per capita affect the
elasticity a lot:
Variable
Stock market partic.
GDP per capita
Credit availability
Real interest
Rule of law
Maximum effect
Std. dev. effect
0.931
0.683
-0.119
-0.265
-0.087
0.141
0.088
-0.020
-0.019
-0.012
DYME, 2 June 2014
21 / 23
Motivation
Methodology
Results
Summary
Results. . . Once Again
Main Findings
1
The elasticity of intertemporal substitution varies a lot
across countries.
2
Households in rich countries and countries with high stock
market participation substitute more intertemporally.
3
Some method choices have systematic effects on the
estimated elasticity (data frequency, inclusion of durables).
Project Website
www.meta-analysis.cz/substitution
DYME, 2 June 2014
22 / 23
Motivation
Methodology
Results
Summary
For Further Reading
Stanley, T. D. & C. Doucouliagos (2012): Meta-Regression
Analysis in Economics and Business.
Routledge, 1st. edition.
Chetty, R., A. Guren, D. Manoli, & A. Weber (2013): Does
Indivisible Labor Explain the Difference between Micro and
Macro Elasticities? A Meta-Analysis.
NBER Macroeconomics Annual 2013: pp. 1–56.
Card, D., J. Kluve, & A. Weber (2010): Active Labour
Market Policy Evaluations: A Meta-Analysis.
The Economic Journal 120(548): pp. F452–F477.
Reading list on RePEc: Google “meta-analysis in economics.”
DYME, 2 June 2014
23 / 23

Cross-Country Heterogeneity in Intertemporal Substitution

Transcription

Similar documents