facts, issues and problems about consumption in nta an-chi tung, 2007.11.05 institute of economics,...
TRANSCRIPT
Facts, Issues and Problems about Consumption in NTA
An-Chi Tung, 2007.11.05Institute of Economics, Academia Sinica
Seoul, 2007.11 NTA Workshop 2
Outline
• Cross-section: 15 countries– shape: why differs from past studies?
– composition: CG and CF; durables; health and education,…
• Time-series//cohort pattern: Taiwan, 1978-2003
– growth effect
– time effect/institutional change
Seoul, 2007.11 NTA Workshop 4
Life cycle hypothesis1. (private) consumption smoothing over the life cycle, based o
n utility maximization (Modigliani and Brumberg, 1954,…)
2. empirical results are mixed• macro (Campbell and Mankiw, 1989,… ) • micro – cross section, panel, or cohort (Hall and Mishkin, 1982;
Altonji and Siow, 1987)
3. micro studies• usually based on household heads at age 20 and above;• some find a hump shape (and the retirement consumption puzzle):
with peak at middle age • due to liquidity constraint (Thurow, 1969), family size (Irvine, 19
78), bequest (Barro, 1974),…
Seoul, 2007.11 NTA Workshop 6
Consumption in NTA
• ex-post accounting, no utility maximization
• estimated by individual, not by household
• Includes both private and public C
• Includes both durable and nondurable C• …
Seoul, 2007.11 NTA Workshop 8
0
5,000
10,000
15,000
20,000
25,000
30,000
35,000
40,000
USA 2003
Franc
e 200
1
Sweden
2003
Austri
a 200
0
Japa
n 200
4
Slove
nia 20
04
Taiwan
1998
S. Kor
ea 20
00
Chile
1997
Urugu
ay 19
94
Costa
Rica 20
04
Thaila
nd 20
03
Philip
pine
s 199
9
Indo
nesia
1999
Indi
a 199
9
Real GDP pc(in 2000 international dollar, from Penn World Table
6.2)
Seoul, 2007.11 NTA Workshop 9
By real GDP pc
US France Sweden Austria Japan
Slovenia Taiwan S. Korea Chile Uruguay
Costa Rica Thailand Philippines Indonesia India
10
Patterns of mean consumption
• Hump shape– Chile 1997 (Indonesia has an esrly hump)
• Rising at old age– US 2003, Sweden 2003, Japan 2004, Uruguay 1994 (India 1999)
• Double hump– France 2001, Costa Rica 2004, South Korea 2000
• Flat during adulthood– Flat after around age 20: Thailand 2004, India 1999
– Flat after fast drop at around age 20: Slovenia 2004, Taiwan 1998
• Other…
Seoul, 2007.11 NTA Workshop 12
Single Hump
conform to Life-cycle hypothesis?
age: 47
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0 10 20 30 40 50 60 70 80 90
Chile 1997
Seoul, 2007.11 NTA Workshop 13
Double hump
“education” peak in Korea and France
CFX shows double hump in Costa Rica
cf. Chayanov cycle: there is a cycle in C with peak at age 25, 55, and perhaps 85, due to change in family size (Mason and Miller, 2000)
C: 27C: 57
F: 53
F: 15 K: 57
K: 18
0.0
0.2
0.4
0.6
0.8
1.0
0 10 20 30 40 50 60 70 80 90
Costa Rica 2004
France 2001
South Korea 2000
Seoul, 2007.11 NTA Workshop 14
Rising at old age
- may also have hump
US: 17J: 17
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
0 10 20 30 40 50 60 70 80 90
Uruguay 1994
US 2003
Japan 2004
Sweden 2003
Seoul, 2007.11 NTA Workshop 15
Flat for adults- some have an
“education” peak
Th: 21
T: 25
T: 19
S: 20
S: 14
0.0
0.2
0.4
0.6
0.8
0 10 20 30 40 50 60 70 80 90
Thailand 2004
India 1999
Taiwan 1998
Slovenia 2004
Seoul, 2007.11 NTA Workshop 16
other typea hump, but peak early
22 28
0.0
0.2
0.4
0.6
0.8
0 10 20 30 40 50 60 70 80 90
Indonesia 1999
Seoul, 2007.11 NTA Workshop 17
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0 10 20 30 40 50 60 70 80 90
high
middle
low
Mean Consumptionby income group (simple average)
as Ron Lee has pointed out:
In the low group, investment in human capital is low, except for Thailand
In the high group, elderly medical expenses are high
In the middle and low groups, adult consumption is flat, except for Urugay.
Seoul, 2007.11 NTA Workshop 18
Why differ from past findings
1. income level or other factors matter:• C on elderly is high in rich countries, and is mostly due to
medical expenditures; but Uruguay is high, too.
• C on children is low in low-income countries, except for Thailand
2. NTA methods to allocate to individuals
3. NTA inclusion of public C and durables
4. data quality or other problems…
Seoul, 2007.11 NTA Workshop 20
Components of Consumption
• C = CG + CF
CG = CGE + CGH + CGX
CF = CFE + CFH + CFD + CFR + CFX
Seoul, 2007.11 NTA Workshop 21
US France Sweden Austria Japan
Slovenia Taiwan S. Korea Chile Uruguay
Costa Rica Thailand Philippines Indonesia India
Seoul, 2007.11 NTA Workshop 22
Three issues: results and problems
1. Inclusion of CG
2. “Consumption” or “C-expenditures”:
- durables vs nondurables
3. What is in CFX (other private consumption)? - In Japan, pocket money, “miscellaneous”, and remittances have double humps, other C’s do not.
- In many economies, durables (CFD) or housing (CFR) are not separately measured.
Seoul, 2007.11 NTA Workshop 23
Public Consumption
CG = CGE + CGH + CGX age-specific pattern per capita
Seoul, 2007.11 NTA Workshop 24
An observation
• Richer countries have larger public consumption expenditures, on both health and education.
Seoul, 2007.11 NTA Workshop 25
US France Sweden Austria Japan
Slovenia Taiwan S. Korea Chile Uruguay
Costa Rica Thailand Philippines Indonesia
Seoul, 2007.11 NTA Workshop 26
Two kinds of public consumption
• Public goods: with non-rivalness and non-excludability, e.g., defense, law,…
→ allocated equally on everyone
• “Merit goods” (or “social goods”,…): goods that could be delivered as private goods, but are delivered by government to avoid under-provision (due to externality or market failure,…), e.g, education, health
→ shows distinct age profile, and estimated separately
Seoul, 2007.11 NTA Workshop 27
A question
• Addibility of “public goods” to private goods?– less problematic with “merit goods”, but
– public consumption is sometimes found to enter into private utility function negatively (Bailey, 1971)
– people derive utility from the “services” of government purchases
• NTA: ex-post accounting– In any case, need to be careful in adding CG to CF
Seoul, 2007.11 NTA Workshop 28
Private Consumption
CF = CFE + CFH + CFD + CFR + CFX no data sometimes
Seoul, 2007.11 NTA Workshop 29
US France Sweden Austria Japan
Slovenia Taiwan S. Korea Chile Uruguay
Costa Rica Thailand Philippines Indonesia India
Seoul, 2007.11 NTA Workshop 30
Mean CF (private consumption)
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0 10 20 30 40 50 60 70 80 90
Austria, 2000
Chile, 1997
Costa Rica, 2004
France, 2001
India, 1999
Indonesia, 1999
Japan, 2004
Philippines, 1999
South Korea, 2000
Slovenia, 2004
Sweden, 2003
Taiwan, 1998
Thailand, 2004
Uruguay, 1994
US, 2003
Seoul, 2007.11 NTA Workshop 31
Mean private consumption
• Most countries show single or double hump shape in CF, e.g., – Sweden: rising at old age → double hump– Austria: flat → hump,…
• Uruguay: CFX dominates all else (as in many other economies)
Seoul, 2007.11 NTA Workshop 32
Uruguay, 1994
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0 10 20 30 40 50 60 70 80 90
CFX
CFH
CFE
Seoul, 2007.11 NTA Workshop 33
CFX (other consumption)
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0 10 20 30 40 50 60 70 80 90
Austria, 2000
Chile, 1997
Costa Rica, 2004
France, 2001
India, 1999
Indonesia, 1999
Japan, 2004
Philippines, 1999
South Korea, 2000
Slovenia, 2004
Sweden, 2003
Taiwan, 1998
Thailand, 2004
Uruguay, 1994
US, 2003
Seoul, 2007.11 NTA Workshop 35
CFD (durables)
• In the literature on life-cycle consumption, the “services” of durables are either ignored or assumed to be additive to nondurable good consumption.
• In NTA, C expenditure on durables are counted as current consumption,– may need to refine.
Seoul, 2007.11 NTA Workshop 36
CFD of 4 countries:
• Japan (29, 59) and Philippines (26, 54) show double hump,
• US has a late peak (58), and Sweden somewhat earlier (48).
Seoul, 2007.11 NTA Workshop 37
CFD (durables)
J, 59
J: 29
Ph, 26Ph, 54
S: 48
A: 58
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0 10 20 30 40 50 60 70 80 90
Japan 2004
Philippines 1999
Sweden 2003
US 2003
Seoul, 2007.11 NTA Workshop 38
• CFD (durable) and CFR (housing) are not calculated in all economies
– due to lack of micro data or aggregate control
• Is it a problem if these are combined into CFX?– how to deal with this?
Seoul, 2007.11 NTA Workshop 40
US France Sweden Austria Japan
Slovenia Taiwan S. Korea Chile Uruguay
Costa Rica Thailand Philippines Indonesia
Seoul, 2007.11 NTA Workshop 41
Health and Education
• An observation: – Rich countries spend more on both, and mostly by
government
• Some definition questions:– Is day care or private tutorship considered education?
– Is long-term care included as medical expenditure?
Seoul, 2007.11 NTA Workshop 43
From Gretchen Donehower (Jan, 2007) : Labor Income and Consumption, 1888-2003
Age
Do
llar
s (
US
, 20
00)
1888
0
2000
4000
6000
8000
0 20 40 60 80
Labor Income
Consumption
1917
0
2000
4000
6000
8000
10000
0 20 40 60 80
1935
0
2000
4000
6000
8000
10000
0 20 40 60 80
1960
0
5000
10000
15000
20000
25000
0 20 40 60 80
1981
0
10000
20000
30000
40000
0 20 40 60 80
2003
0
10000
20000
30000
40000
50000
60000
0 20 40 60 80
Seoul, 2007.11 NTA Workshop 44
Taiwan, 1978, 1988, 1998, 2003
Taiwan, 1978
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0 10 20 30 40 50 60 70 80 90
YL
C
CF
CG
Taiwan, 1998
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0 10 20 30 40 50 60 70 80 90
Taiwan, 1988
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0 10 20 30 40 50 60 70 80 90
Taiwan, 2003
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0 10 20 30 40 50 60 70 80 90
Seoul, 2007.11 NTA Workshop 45
(normalized by average YL pc, age 30-49)
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0 10 20 30 40 50 60 70 80 90+
Taiwan 2003Taiwan 1998Taiwan 1991Taiwan 1986Taiwan 1981
Taiwan, 1981-2003
Seoul, 2007.11 NTA Workshop 46
Increase of education and health over time
Mean consumption (normalized)
0%
10%
20%
30%
40%
50%
60%
70%
80%
0-19 20-64 65+
CGE
CGH
CGX
CFE
CFH
CFR
CFX
0%
10%
20%
30%
40%
50%
60%
70%
80%
0-19 20-64 65+
CGE
CGH
CGX
CFE
CFH
CFR
CFX
0-19 20-64 65+ 0-19 20-64 65+
1978 1998
Seoul, 2007.11 NTA Workshop 48
Mean Real Consumptiondeflated by GDP deflator
0
20000
40000
60000
80000
100000
120000
140000
0 10 20 30 40 50 60 70 80 90
0
50000
100000
150000
200000
250000
300000
0 10 20 30 40 50 60 70 80 90
C1998
C1983
C1978
C1958 C1938 C1918 C18980
50000
100000
150000
200000
250000
300000
350000
400000
450000
0 10 20 30 40 50 60 70 80 90
C1998 C1988 C1983 C1978 C1968 C1958
C1948 C1938 C1928 C1918 C1908 C1898
CF
CG
Seoul, 2007.11 NTA Workshop 49
Growth in mean LY
Real (deflated by GDP deflator) normalized by average
YL 30-49
C1958, 42
C1948, 48
C1938, 54
0
100000
200000
300000
400000
500000
600000
0 10 20 30 40 50 60 70 80 90
C1998
C1988
C1978
C1968
C1958
C1948
C1938
C1928
C1918
C1908
C1898
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0 10 20 30 40 50 60 70 80 90
Seoul, 2007.11 NTA Workshop 50
Taiwan’s real GDP pc (in international $)
20,701
3,755
0
5000
10000
15000
20000
25000
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
Seoul, 2007.11 NTA Workshop 51
Normalized mean consumption
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
0 10 20 30 40 50 60 70 80 90
C1998 C1988 C1983 C1978 C1968 C1958
C1948 C1938 C1928 C1918 C1908 C1898
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0 10 20 30 40 50 60 70 80 90
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0 10 20 30 40 50 60 70 80 90
CF
CG
Seoul, 2007.11 NTA Workshop 52
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
0 10 20 30 40 50 60 70 80 90
C1998 C1988 C1983 C1978 C1968 C1958
C1948 C1938 C1928 C1918 C1908 C1898
(normalized by average YL pc, age 30-49)
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0 10 20 30 40 50 60 70 80 90+
Taiwan 2003Taiwan 1998Taiwan 1991Taiwan 1986Taiwan 1981
“normalized”cohort
time series