1240
Views
545
Downloads |
Education spending, economic development, and the size of government
Article | Year: 2023 | Pages: 285 - 333 | Volume: 47 | Issue: 3 Received: September 29, 2022 | Accepted: June 21, 2023 | Published online: September 4, 2023
|
FULL ARTICLE
FIGURES & DATA
REFERENCES
CROSSMARK POLICY
METRICS
LICENCING
PDF
|
Variable name
|
Description of the variable
|
Source
|
|
Dependent variables
|
|
pse/gdp
|
Public spending on education, total (% of GDP)
|
World Bank EdStats
|
|
pse/gov
|
Public spending on education, total (% of total government spending)
|
World Bank EdStats
|
|
Explanatory variables
|
|
ypc2015
|
GNI per capita country grouping in 2015, 21 OECD countries
|
World Bank (Atlas Method)
|
|
region
|
Richer (versus poorer) country regions
|
Authors’ compilation
|
|
poldemoc
|
Political democracy classification: yes; no
|
Freedom House
|
|
Control variables
|
|
pop024
|
Population aged 0-24 (% of total population)
|
World Bank EdStats
|
|
urban
|
Urban population (% of total population)
|
World Bank WDI
|
|
trade
|
Exports plus imports of goods and services (% of GDP)
|
World Bank WDI
|
|
hci
|
Human capital index
|
Penn World Table 9.0
|
|
pop65
|
Population aged 65 and above (% of total population)
|
World Bank WDI
|
|
military
|
Military expenditure (% of GDP)
|
World Bank WDI
|
|
fiscbal
|
Fiscal balance (% of GDP)
|
World Bank DPG
|
|
debt
|
General government gross debt (IMF, % of GDP)
|
World Bank TCdata360
|
|
Other variables
|
|
gdppc
|
GDP per capita, PPP (constant 2011 international $)
|
World Bank WDI
|
|
gini
|
Gini index (World Bank Estimate)
|
World Bank WDI
|
Notes: EdStats refers to the World Bank’s Education Statistics database (World Bank, 2017a). TCdata360 refers to the World Bank’s TCdata360 database (World Bank, 2017b). WDI refers to the World Bank’s World Development Indicators database (World Bank, 2017c). DPG refers to the World Bank’s Development Prospects Group: A Cross-Country Database of Fiscal Space (World Bank, 2017d). The pop024 variable is the sum of pop014 and pop1524 variables from the World Bank EdStats database. Freedom House refers to the Freedom in the World survey data (Freedom House, 2016). See Feenstra, Inklaar and Timmer (2015) for the Penn World Table 9.0.
|
Dependent variable
|
pse/gdp
|
pse/gov
|
|
Estimation method
|
LSDV
|
Quantile
|
Robust
|
LSDV
|
Quantile
|
Robust
|
|
|
(A)
|
(B)
|
(A)
|
(B)
|
(A)
|
(B)
|
(A)
|
(B)
|
(A)
|
(B)
|
(A)
|
(B)
|
|
1#0. Low income, not democratic
|
-1.551***
|
-1.635***
|
-1.787***
|
-1.885***
|
-1.979***
|
-2.013***
|
3.633***
|
-0.012
|
3.948***
|
-0.162
|
3.387***
|
-0.210
|
|
(0.254)
|
(0.359)
|
(0.133)
|
(0.260)
|
(0.130)
|
(0.207)
|
(0.436)
|
(0.726)
|
(0.608)
|
(0.696)
|
(0.392)
|
(0.614)
|
|
1#1. Low income, democratic
|
-1.684***
|
-1.756***
|
-1.757***
|
-1.877***
|
-1.724***
|
-1.759***
|
4.796***
|
1.339*
|
5.135***
|
1.290**
|
4.821***
|
1.467**
|
|
(0.123)
|
(0.257)
|
(0.127)
|
(0.257)
|
(0.153)
|
(0.216)
|
(0.381)
|
(0.686)
|
(0.366)
|
(0.581)
|
(0.467)
|
(0.643)
|
|
2#0. Lower middle income, not democratic
|
-1.152***
|
-1.412***
|
-1.170***
|
-1.578***
|
-1.362***
|
-1.611***
|
4.304***
|
1.081*
|
4.231***
|
0.614
|
4.196***
|
0.850*
|
|
(0.130)
|
(0.220)
|
(0.209)
|
(0.259)
|
(0.116)
|
(0.171)
|
(0.368)
|
(0.600)
|
(0.595)
|
(0.531)
|
(0.353)
|
(0.511)
|
|
2#1. Lower middle income, democratic
|
-0.454***
|
-0.663***
|
-0.739**
|
-1.126***
|
-0.979***
|
-1.123***
|
4.852***
|
2.421***
|
4.137***
|
1.661***
|
4.224***
|
1.830***
|
|
(0.158)
|
(0.199)
|
(0.318)
|
(0.287)
|
(0.117)
|
(0.159)
|
(0.369)
|
(0.446)
|
(0.470)
|
(0.419)
|
(0.348)
|
(0.468)
|
|
3#0. Upper middle income, not democratic
|
-1.090***
|
-1.389***
|
-1.357***
|
-1.669***
|
-1.432***
|
-1.764***
|
2.520***
|
-0.022
|
2.058***
|
-0.131
|
2.317***
|
-0.128
|
|
(0.155)
|
(0.230)
|
(0.166)
|
(0.177)
|
(0.129)
|
(0.146)
|
(0.455)
|
(0.503)
|
(0.748)
|
(0.371)
|
(0.405)
|
(0.450)
|
|
3#1. Upper middle income, democratic
|
-0.819***
|
-1.097***
|
-0.797***
|
-1.131***
|
-0.935***
|
-1.133***
|
3.643***
|
1.708***
|
3.556***
|
1.205***
|
3.411***
|
1.347***
|
|
(0.097)
|
(0.132)
|
(0.106)
|
(0.156)
|
(0.106)
|
(0.127)
|
(0.279)
|
(0.354)
|
(0.331)
|
(0.179)
|
(0.318)
|
(0.374)
|
|
4#0. High income (non-OECD), not democratic
|
-1.123***
|
-1.987***
|
-1.478***
|
-1.933***
|
-1.329***
|
-2.047***
|
0.987**
|
-2.057***
|
0.034
|
-3.107***
|
0.683
|
-2.852***
|
|
(0.183)
|
(0.217)
|
(0.152)
|
(0.195)
|
(0.171)
|
(0.189)
|
(0.499)
|
(0.566)
|
(0.751)
|
(0.356)
|
(0.502)
|
(0.543)
|
|
4#1. High income (non-OECD), democratic
|
-0.814***
|
-0.768***
|
-0.644***
|
-0.654***
|
-0.777***
|
-0.721***
|
0.568**
|
-0.038
|
0.923***
|
0.308
|
0.471
|
-0.099
|
|
(0.091)
|
(0.101)
|
(0.123)
|
(0.130)
|
(0.106)
|
(0.111)
|
(0.249)
|
(0.254)
|
(0.308)
|
(0.342)
|
(0.333)
|
(0.338)
|
|
5#0. High income (OECD), not democratic
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
|
5#1. High income (OECD), democratic BASE
|
5.352***
|
3.668***
|
5.226***
|
3.107***
|
5.325***
|
3.449***
|
12.245***
|
4.332***
|
11.973***
|
0.885
|
12.217***
|
2.576***
|
|
(0.055)
|
(0.312)
|
(0.056)
|
(0.334)
|
(0.073)
|
(0.260)
|
(0.122)
|
(0.831)
|
(0.166)
|
(0.643)
|
(0.214)
|
(0.768)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
pop024
|
|
0.016***
|
|
0.022***
|
|
0.018***
|
|
0.146***
|
|
0.191***
|
|
0.164***
|
|
|
(0.006)
|
|
(0.006)
|
|
(0.004)
|
|
(0.015)
|
|
(0.010)
|
|
(0.013)
|
|
urban
|
|
0.009***
|
|
0.011***
|
|
0.011***
|
|
0.029***
|
|
0.046***
|
|
0.040***
|
|
|
(0.002)
|
|
(0.003)
|
|
(0.002)
|
|
(0.007)
|
|
(0.006)
|
|
(0.006)
|
|
trade
|
|
0.008***
|
|
0.008***
|
|
0.007***
|
|
0.017***
|
|
0.022***
|
|
0.021***
|
|
|
(0.001)
|
|
(0.001)
|
|
(0.001)
|
|
(0.003)
|
|
(0.003)
|
|
(0.002)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
R-squared
|
0.060
|
0.100
|
0.074
|
0.107
|
0.123
|
0.190
|
0.141
|
0.198
|
0.092
|
0.146
|
0.136
|
0.226
|
|
F-value
|
35.86***
|
34.89***
|
n/a
|
n/a
|
43.22***
|
48.51***
|
64.55***
|
56.40***
|
n/a
|
n/a
|
42.96***
|
54.58***
|
|
Countries
|
183
|
169
|
183
|
169
|
183
|
169
|
175
|
165
|
175
|
165
|
175
|
165
|
|
Years
|
27
|
26
|
27
|
26
|
27
|
26
|
27
|
26
|
27
|
26
|
27
|
26
|
|
Observations
|
2468
|
2288
|
2468
|
2288
|
2468
|
2288
|
2194
|
2069
|
2194
|
2069
|
2194
|
2069
|
Notes: BASE group is high-income (OECD) and democratic countries. A pseudo R-squared is reported for the quantile regression. Not applicable (n/a) means there are no applicable countries for these groups. Standard errors are given in parentheses. Huber/White heteroskedasticity-robust standard errors are reported for the LSDV and quantile estimators. Significance levels: * p < 0.10, ** p < 0.05, *** p < 0.01.
|
Dependent variable
|
pse/gdp
|
pse/gov
|
|
Estimation method
|
LSDV
|
Quantile
|
Robust
|
LSDV
|
Quantile
|
Robust
|
|
|
(A)
|
(B)
|
(A)
|
(B)
|
(A)
|
(B)
|
(A)
|
(B)
|
(A)
|
(B)
|
(A)
|
(B)
|
|
0#0. Poorer country regions, not democratic
|
-1.298***
|
-2.031***
|
-1.633***
|
-1.756***
|
-1.681***
|
-1.760***
|
3.880***
|
1.492**
|
4.046***
|
0.726
|
3.757***
|
1.108*
|
|
(0.147)
|
(0.216)
|
(0.117)
|
(0.234)
|
(0.104)
|
(0.199)
|
(0.305)
|
(0.612)
|
(0.499)
|
(0.764)
|
(0.303)
|
(0.597)
|
|
0#1. Poorer country regions, democratic
|
-0.804***
|
-1.412***
|
-1.163***
|
-1.235***
|
-1.110***
|
-1.178***
|
5.077***
|
3.321***
|
4.833***
|
2.284***
|
4.923***
|
2.945***
|
|
(0.104)
|
(0.166)
|
(0.121)
|
(0.180)
|
(0.098)
|
(0.166)
|
(0.233)
|
(0.458)
|
(0.287)
|
(0.496)
|
(0.283)
|
(0.491)
|
|
1#0. Richer country regions, not democratic
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
|
1#1. Richer country regions, democratic BASE
|
5.169***
|
2.095***
|
5.114***
|
2.047***
|
5.213***
|
2.520***
|
11.944***
|
5.333***
|
11.849***
|
2.093**
|
11.943***
|
3.705***
|
|
(0.069)
|
(0.375)
|
(0.062)
|
(0.404)
|
(0.074)
|
(0.348)
|
(0.127)
|
(1.193)
|
(0.160)
|
(0.960)
|
(0.212)
|
(1.036)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
pop024
|
|
0.034***
|
|
0.025***
|
|
0.019***
|
|
0.104***
|
|
0.163***
|
|
0.127***
|
|
|
(0.007)
|
|
(0.007)
|
|
(0.006)
|
|
(0.021)
|
|
(0.019)
|
|
(0.019)
|
|
urban
|
|
0.016***
|
|
0.022***
|
|
0.020***
|
|
0.029***
|
|
0.044***
|
|
0.037***
|
|
|
(0.003)
|
|
(0.003)
|
|
(0.002)
|
|
(0.008)
|
|
(0.008)
|
|
(0.007)
|
|
trade
|
|
0.013***
|
|
0.011***
|
|
0.011***
|
|
0.018***
|
|
0.017***
|
|
0.021***
|
|
|
(0.002)
|
|
(0.002)
|
|
(0.001)
|
|
(0.004)
|
|
(0.004)
|
|
(0.003)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
R-squared
|
0.057
|
0.167
|
0.090
|
0.170
|
0.154
|
0.313
|
0.195
|
0.222
|
0.140
|
0.171
|
0.189
|
0.247
|
|
F-value
|
52.61***
|
63.60***
|
n/a
|
n/a
|
135.10***
|
125.25***
|
270.00***
|
113.21***
|
n/a
|
n/a
|
158.19***
|
84.82***
|
|
Countries
|
102
|
97
|
102
|
97
|
102
|
97
|
99
|
96
|
99
|
96
|
99
|
96
|
|
Years
|
27
|
26
|
27
|
26
|
27
|
26
|
27
|
26
|
27
|
26
|
27
|
26
|
|
Observations
|
1486
|
1382
|
1486
|
1382
|
1486
|
1382
|
1360
|
1299
|
1360
|
|
1360
|
1299
|
Notes: BASE group is richer country regions that are democratic. See the notes for table 2. Standard errors are given in parentheses. Significance levels: * p < 0.10, ** p < 0.05, *** p < 0.01.
|
Wald tests for parameter
equality in table 2
|
|
|
Robust estimator
|
|
(A)
|
|
(B)
|
|
Dependent variable: pse/gdp
|
Wald Statistic
|
Uncorrected
Significance
(α = 0.05)
|
Corrected
Significance
(α = 0.0125)
|
Wald Statistic
|
Uncorrected
Significance
(α = 0.05)
|
Corrected
Significance
(α = 0.0125)
|
|
Test parameter 1#0 = 1#1
|
F(1, 2459)
= 2.19
|
No
|
No
|
F (1, 2276) = 2.33
|
No
|
No
|
|
p = 0.1388
|
p = 0.1273
|
|
Test parameter 2#0 = 2#1
|
F(1, 2459) =
8.90
|
Yes
|
Yes
|
F(1, 2276) = 14.79
|
Yes
|
Yes
|
|
p = 0.0029
|
p = 0.0001
|
|
Test parameter 3#0 = 3#1
|
F(1, 2459)
= 14.38
|
Yes
|
Yes
|
F(1, 2276) = 24.17
|
Yes
|
Yes
|
|
p = 0.0002
|
p = 0.0000
|
|
Test parameter 4#0 = 4#1
|
F(1, 2459)
= 10.20
|
Yes
|
Yes
|
F(1, 2276) = 51.44
|
Yes
|
Yes
|
|
p = 0.0014
|
p = 0.0000
|
|
|
Robust estimator
|
|
(A)
|
(B)
|
|
Dependent variable: pse/gov
|
Wald Statistic
|
Uncorrected
Significance
(α = 0.05)
|
Corrected
Significance
(α = 0.0125)
|
Wald Statistic
|
Uncorrected
Significance
(α = 0.05)
|
Corrected
Significance
(α = 0.0125)
|
|
Test parameter 1#0 = 1#1
|
F(1, 2185) = 7.33
|
Yes
|
Yes
|
F(1, 2057)
= 11.37
|
Yes
|
Yes
|
|
p = 0.0068
|
p = 0.0008
|
|
Test parameter 2#0 = 2#1
|
F(1, 2185)
= 0.00
|
No
|
No
|
F(1, 2057)
= 6.75
|
Yes
|
Yes
|
|
p = 0.9438
|
p = 0.0094
|
|
Test parameter 3#0 = 3#1
|
F(1, 2185)
= 6.88
|
Yes
|
Yes
|
F(1, 2057)
= 13.88
|
Yes
|
Yes
|
|
p = 0.0088
|
p = 0.0002
|
|
Test parameter 4#0 = 4#1
|
F(1, 2185)
= 0.16
|
No
|
No
|
F(1, 2057)
= 25.58
|
Yes
|
Yes
|
|
p = 0.6852
|
p = 0.0000
|
|
Wald tests for parameter
equality in table 3
|
|
|
Robust estimator
|
|
(A)
|
(B)
|
|
Dependent variable: pse/gdp
|
Wald Statistic
|
Uncorrected
Significance
(α = 0.05)
|
Corrected
Significance
(α = 0.05)
|
Wald Statistic
|
Uncorrected
Significance
(α = 0.05)
|
Corrected
Significance
(α = 0.05)
|
|
Test parameter 0#0 = 0#1
|
F(1, 1483)
= 33.66
|
Yes
|
Yes
|
F(1, 1376)
= 35.80
|
Yes
|
Yes
|
|
p = 0.0000
|
p = 0.0000
|
|
|
Robust estimator
|
|
(A)
|
(B)
|
|
Dependent variable: pse/gov
|
Wald Statistic
|
Uncorrected
Significance
(α = 0.05)
|
Corrected
Significance
(α = 0.05)
|
Wald Statistic
|
Uncorrected
Significance
(α = 0.05)
|
Corrected
Significance
(α = 0.05)
|
|
Test parameter 0#0 = 0#1
|
F(1, 1357)
= 16.59
|
Yes
|
Yes
|
F(1, 1293)
= 37.70
|
Yes
|
Yes
|
|
p = 0.0000
|
p = 0.0000
|
Notes: E refers to public spending on education, Y to national income (GDP) and G to total public spending. Subscripts R and P refer to richer and poorer countries, respectively. If Propositions 1 and 2 hold true, then they imply Proposition 3.
|
Low income
(31)
|
Lower middle income
(52)
|
Upper middle income
(56)
|
High income
(mostly non-OECD) (57)
|
High income
(OECD) (21)
|
|
Afghanistan
Benin
Burkina Faso
Burundi
Central African Republic
Chad
Comoros
Congo, Dem. Rep.
Eritrea
Ethiopia
Gambia, The
Guinea
Guinea-Bissau
Haiti
Korea, Dem. People’s Rep.
Liberia
Madagascar
Malawi
Mali
Mozambique
Nepal
Niger
Rwanda
Senegal
Sierra Leone
Somalia
South Sudan
Tanzania
Togo
Uganda
Zimbabwe
|
Armenia
Bangladesh
Bhutan
Bolivia
Cabo Verde
Cambodia
Cameroon
Congo, Rep.
Cote d'Ivoire
Djibouti
Egypt, Arab Rep.
El Salvador
Ghana
Guatemala
Honduras
India
Indonesia
Kenya
Kiribati
Kosovo
Kyrgyz Republic
Lao PDR
Lesotho
Mauritania
Micronesia, Fed. Sts.
Moldova
Mongolia
Morocco
Myanmar
Nicaragua
Nigeria
Pakistan
Papua New Guinea
Philippines
Samoa
Sao Tome and Principe
Solomon Islands
Sri Lanka
Sudan
Swaziland
Syrian Arab Republic
Tajikistan
Timor-Leste
Tonga
Tunisia
Ukraine
Uzbekistan
Vanuatu
Vietnam
West Bank and Gaza
Yemen, Rep.
Zambia
|
Albania
Algeria
American Samoa
Angola
Argentina
Azerbaijan
Belarus
Belize
Bosnia and Herzegovina
Botswana
Brazil
Bulgaria
China
Colombia
Costa Rica
Cuba
Dominica
Dominican Republic
Ecuador
Equatorial Guinea
Fiji
Gabon
Georgia
Grenada
Guyana
Iran, Islamic Rep.
Iraq
Jamaica
Jordan
Kazakhstan
Lebanon
Libya
Macedonia, FYR
Malaysia
Maldives
Marshall Islands
Mauritius
Mexico
Montenegro
Namibia
Palau
Panama
Paraguay
Peru
Romania
Russian Federation
Serbia
South Africa
St. Lucia
St. Vincent and the Grenadines
Suriname
Thailand
Turkey
Turkmenistan
Tuvalu
Venezuela, RB
|
Andorra
Antigua and Barbuda
Aruba
Bahamas, The
Bahrain
Barbados
Bermuda
British Virgin Islands
Brunei Darussalam
Cayman Islands
Channel Islands
Chile
Croatia
Curacao
Cyprus
Czech Republic
Estonia
Faroe Islands
French Polynesia
Gibraltar
Greenland
Guam
Hong Kong SAR, China
Hungary
Iceland
Isle of Man
Israel
Korea, Rep.
Kuwait
Latvia
Liechtenstein
Lithuania
Luxembourg
Macao SAR, China
Malta
Monaco
Nauru
New Caledonia
Northern Mariana Islands
Oman
Poland
Puerto Rico
Qatar
San Marino
Saudi Arabia
Seychelles
Singapore
Sint Maarten (Dutch part)
Slovak Republic
Slovenia
St. Kitts and Nevis
St. Martin (French part)
Trinidad and Tobago
Turks and Caicos Islands
United Arab Emirates
Uruguay
Virgin Islands (U.S.)
|
Australia
Austria
Belgium
Canada
Denmark
Finland
France
Germany
Greece
Ireland
Italy
Japan
Netherlands
New Zealand
Norway
Portugal
Spain
Sweden
Switzerland
United Kingdom
United States
|
Notes: Groups are adapted from the World Bank’s Country and Lending Groups for the 2015 calendar year, based on GNI per capita calculated using the World Bank Atlas Method, except for the high-income (OECD) group, which includes the 21 countries comprising the “core” OECD countries that have been categorised as OECD for the entire study period, from 1989 to 2015 (i.e., excluding Chile, Czech Republic, Estonia, Hungary, Iceland, Israel, Korea, Latvia, Luxembourg, Mexico, Poland, Slovak Republic, Slovenia and Turkey, which are mostly included in the second high-income group). Numbers in parentheses show the total number of countries in each group. The historical classification is available from: https://datahelpdesk.worldbank.org/knowledgebase/articles/906519-world-bank-country-and-lending-groups. Source: Adapted from the World Bank’s historical classification.
|
Poorer country regions
|
|
Central Africa (8)
|
Central America (8)
|
East Africa (12)
|
South America (12)
|
South Asia (8)
|
|
Cameroon
Central African Republic
Chad
Congo, Dem. Rep.
Congo, Rep.
Equatorial Guinea
Gabon
Sao Tome and Principe
|
Belize
Costa Rica
El Salvador
Guatemala
Honduras
Mexico
Nicaragua
Panama
|
Burundi
Comoros
Djibouti
Eritrea
Ethiopia
Kenya
Rwanda
Somalia
South Sudan
Sudan
Tanzania
Uganda
|
Argentina
Bolivia
Brazil
Chile
Colombia
Ecuador
Guyana
Paraguay
Peru
Suriname
Uruguay
Venezuela, RB
|
Afghanistan
Bangladesh
Bhutan
India
Maldives
Nepal
Pakistan
Sri Lanka
|
|
Southeast Asia (11)
|
Southern Africa (13)
|
West Africa (16)
|
|
|
|
Brunei Darussalam
Cambodia
Indonesia
Lao PDR
Malaysia
Myanmar
Philippines
Singapore
Thailand
Timor-Leste
Vietnam
|
Angola
Botswana
Lesotho
Madagascar
Malawi
Mauritius
Mozambique
Namibia
Seychelles
South Africa
Swaziland
Zambia
Zimbabwe
|
Benin
Burkina Faso
Cabo Verde
Cote d’Ivoire
Gambia, The
Ghana
Guinea
Guinea-Bissau
Liberia
Mali
Mauritania
Niger
Nigeria
Senegal
Sierra Leone
Togo
|
|
|
|
Richer country regions
|
|
North America (3)
|
Nordic Countries (5)
|
Western Europe (22)
|
|
|
|
Bermuda
Canada
United States
|
Denmark
Finland
Iceland
Norway
Sweden
|
Andorra
Austria
Belgium
Channel Islands
Faroe Islands
France
Germany
Gibraltar
Greece
Greenland
Ireland
Isle of Man
Italy
Liechtenstein
Luxembourg
Monaco
Netherlands
Portugal
San Marino
Spain
Switzerland
United Kingdom
|
|
|
Notes: The numbers in parentheses show the total number of countries in each sub-group of the respective country regions. For the poorer country regions, Equatorial Guinea, Chile, Uruguay, Brunei Darussalam, Singapore and Seychelles (the countries in bold) are excluded because they are classified as high-income (non-OECD) countries for most or all of the time period under investigation (from 1989 to 2015). Source: Authors’ compilation.
|
Variable name
|
Data coverage
|
N
|
Countries
|
Years
|
Mean
|
Std dev.
|
Minimum
|
Maximum
|
|
pse/gdp
|
1989-2015
|
2551
|
193
|
13.2
|
4.505
|
2.007
|
0.781
|
44.334
|
|
pse/gov
|
1989-2015
|
2255
|
181
|
12.5
|
14.849
|
5.036
|
2.563
|
47.279
|
|
ypc2015
|
1989-2015
|
5859
|
217
|
27.0
|
n/a
|
n/a
|
n/a
|
n/a
|
|
region
|
1989-2015
|
3024
|
112
|
27.0
|
n/a
|
n/a
|
n/a
|
n/a
|
|
poldemoc
|
1989-2015
|
5105
|
193
|
26.5
|
n/a
|
n/a
|
n/a
|
n/a
|
|
pop024
|
1990-2015
|
4714
|
184
|
25.6
|
49.977
|
13.687
|
20.160
|
73.288
|
|
urban
|
1989-2015
|
5799
|
215
|
27.0
|
55.788
|
24.901
|
5.342
|
100.000
|
|
trade
|
1989-2015
|
4785
|
193
|
24.8
|
86.996
|
52.290
|
0.021
|
531.737
|
|
hci
|
1989-2014
|
3703
|
143
|
25.9
|
2.342
|
0.694
|
1.028
|
3.734
|
|
pop65
|
1989-2015
|
5234
|
195
|
26.8
|
7.073
|
4.814
|
0.697
|
26.342
|
|
military
|
1989-2015
|
3870
|
166
|
23.3
|
2.433
|
3.210
|
0
|
117.388
|
|
fiscbal
|
1990-2015
|
4184
|
191
|
21.9
|
-2.299
|
13.715
|
-505.442
|
122.188
|
|
debt
|
1989-2015
|
3796
|
186
|
20.4
|
57.015
|
49.714
|
0
|
789.833
|
|
gini
|
1989-2014
|
1188
|
155
|
7.7
|
39.875
|
9.871
|
16.23
|
65.76
|
|
gdppc
|
1990-2015
|
4803
|
195
|
24.6
|
15111
|
18507
|
247
|
137164
|
Notes: Years refers to the average number of years (time-series observations) for each country. Std. dev. refers to the overall standard deviation. Two changes were made to the original data for the pse/gdp variable. The zero observation for Turkey in 1998 was deleted (because there were no other 0% values in the dataset; nil or negligible appeared in the original UNESCO source data for this observation) and the observation for Tuvalu in 1997 (3730833.5%) was deleted as an obvious mistake; the extreme value for this observation also appeared in the original UNESCO source data. It was subsequently noted that this observation was deleted from the World Bank’s Education Statistics (EdStats) data as of the update dated 21 May 2018. Descriptive results are not reported for ypc2015 (21 OECD countries), region and poldemoc because these are sets of binary variables used to characterise broad political and economic categories.
|
Dependent variable: pse/gdp
|
LSDV
|
LSDV
(One-way; Country)
|
LSDV
(One-way; Year)
|
LSDV
(Two-way)
|
Newey-West
|
Driscoll-Kraay
|
|
(A)
|
(B)
|
(A)
|
(B)
|
(A)
|
(B)
|
(A)
|
(B)
|
(A)
|
(B)
|
(A)
|
(B)
|
|
1#0. Low income, not democratic
|
-1.551***
|
-1.635***
|
-1.551***
|
-1.635*
|
-1.551***
|
-1.635***
|
-1.551***
|
-1.635*
|
-1.551***
|
-1.635***
|
-1.551***
|
-1.635***
|
|
(0.254)
|
(0.359)
|
(0.544)
|
(0.911)
|
(0.265)
|
(0.394)
|
(0.549)
|
(0.926)
|
(0.311)
|
(0.486)
|
(0.294)
|
(0.531)
|
|
1#1. Low income, democratic
|
-1.684***
|
-1.756***
|
-1.684***
|
-1.756**
|
-1.684***
|
-1.756***
|
-1.684***
|
-1.756**
|
-1.684***
|
-1.756***
|
-1.684***
|
-1.756***
|
|
(0.123)
|
(0.257)
|
(0.330)
|
(0.774)
|
(0.129)
|
(0.255)
|
(0.332)
|
(0.773)
|
(0.178)
|
(0.389)
|
(0.185)
|
(0.371)
|
|
2#0. Lower middle income, not democratic
|
-1.152***
|
-1.412***
|
-1.152***
|
-1.412**
|
-1.152***
|
-1.412***
|
-1.152***
|
-1.412**
|
-1.152***
|
-1.412***
|
-1.152***
|
-1.412***
|
|
(0.130)
|
(0.220)
|
(0.406)
|
(0.688)
|
(0.108)
|
(0.197)
|
(0.399)
|
(0.681)
|
(0.196)
|
(0.338)
|
(0.152)
|
(0.290)
|
|
2#1. Lower middle income, democratic
|
-0.454***
|
-0.663***
|
-0.454
|
-0.663
|
-0.454***
|
-0.663***
|
-0.454
|
-0.663
|
-0.454*
|
-0.663**
|
-0.454***
|
-0.663***
|
|
(0.158)
|
(0.199)
|
(0.484)
|
(0.627)
|
(0.108)
|
(0.131)
|
(0.470)
|
(0.609)
|
(0.235)
|
(0.305)
|
(0.162)
|
(0.169)
|
|
3#0. Upper middle income, not democratic
|
-1.090***
|
-1.389***
|
-1.090**
|
-1.389*
|
-1.090***
|
-1.389***
|
-1.090**
|
-1.389*
|
-1.090***
|
-1.389***
|
-1.090***
|
-1.389***
|
|
(0.155)
|
(0.230)
|
(0.534)
|
(0.801)
|
(0.131)
|
(0.140)
|
(0.527)
|
(0.780)
|
(0.244)
|
(0.369)
|
(0.195)
|
(0.193)
|
|
3#1. Upper middle income, democratic
|
-0.819***
|
-1.097***
|
-0.819***
|
-1.097**
|
-0.819***
|
-1.097***
|
-0.819***
|
-1.097***
|
-0.819***
|
-1.097***
|
-0.819***
|
-1.097***
|
|
(0.097)
|
(0.132)
|
(0.295)
|
(0.423)
|
(0.085)
|
(0.119)
|
(0.292)
|
(0.419)
|
(0.141)
|
(0.203)
|
(0.092)
|
(0.159)
|
|
4#0. High income (non-OECD), not democratic
|
-1.123***
|
-1.987***
|
-1.123**
|
-1.987***
|
-1.123***
|
-1.987***
|
-1.123**
|
-1.987***
|
-1.123***
|
-1.987***
|
-1.123***
|
-1.987***
|
|
(0.183)
|
(0.217)
|
(0.514)
|
(0.702)
|
(0.197)
|
(0.192)
|
(0.519)
|
(0.695)
|
(0.249)
|
(0.332)
|
(0.273)
|
(0.259)
|
|
4#1. High income (non-OECD), democratic
|
-0.814***
|
-0.768***
|
-0.814**
|
-0.768**
|
-0.814***
|
-0.768***
|
-0.814**
|
-0.768**
|
-0.814***
|
-0.768***
|
-0.814***
|
-0.768***
|
|
(0.091)
|
(0.101)
|
(0.338)
|
(0.357)
|
(0.049)
|
(0.064)
|
(0.329)
|
(0.348)
|
(0.144)
|
(0.156)
|
(0.059)
|
(0.085)
|
|
5#0. High income (OECD), not democratic
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
|
5#1. High income (OECD), democratic BASE
|
5.352***
|
3.668***
|
5.352***
|
3.668***
|
5.352***
|
3.668***
|
5.352***
|
3.668***
|
5.352***
|
3.668***
|
5.352***
|
3.668***
|
|
(0.055)
|
(0.312)
|
(0.225)
|
(0.945)
|
(0.058)
|
(0.171)
|
(0.226)
|
(0.908)
|
(0.089)
|
(0.467)
|
(0.083)
|
(0.200)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
pop024
|
|
0.016***
|
|
0.016
|
|
0.016***
|
|
0.016
|
|
0.016*
|
|
0.016**
|
|
|
(0.006)
|
|
(0.018)
|
|
(0.005)
|
|
(0.018)
|
|
(0.009)
|
|
(0.007)
|
|
urban
|
|
0.009***
|
|
0.009
|
|
0.009***
|
|
0.009
|
|
0.009**
|
|
0.009***
|
|
|
(0.002)
|
|
(0.008)
|
|
(0.001)
|
|
(0.007)
|
|
(0.004)
|
|
(0.002)
|
|
trade
|
|
0.008***
|
|
0.008**
|
|
0.008***
|
|
0.008**
|
|
0.008***
|
|
0.008***
|
|
|
(0.001)
|
|
(0.004)
|
|
(0.001)
|
|
(0.003)
|
|
(0.002)
|
|
(0.001)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
R-squared
|
0.060
|
0.100
|
0.060
|
0.100
|
0.060
|
0.100
|
0.060
|
0.100
|
0.060
|
0.100
|
0.060
|
0.100
|
|
F-value
|
35.86***
|
34.89***
|
3.78***
|
3.79***
|
78.63***
|
120.42***
|
n/a
|
n/a
|
15.83***
|
15.72***
|
60.00***
|
387.40***
|
|
Countries
|
183
|
169
|
183
|
169
|
183
|
169
|
183
|
169
|
183
|
169
|
183
|
169
|
|
Years
|
27
|
26
|
27
|
26
|
27
|
26
|
27
|
26
|
27
|
26
|
27
|
26
|
|
Observations
|
2468
|
2288
|
2468
|
2288
|
2468
|
2288
|
2468
|
2288
|
2468
|
2288
|
2468
|
2288
|
Notes: BASE group is high-income (OECD) and democratic countries. Three controls are used: youth population (pop024); urban population (urban) and trade (trade). A pseudo R-squared is reported for the quantile regression. Not applicable (n/a) means there are no applicable countries for these groups. The LSDV estimator using Huber/White heteroskedasticity-robust standard errors is reproduced from table 2 for ease of comparison. The various other standard error estimates for LSDV use one-way (country or year) and two-way (country and year) cluster-robust standard errors. The Newey-West and Driscoll-Kraay estimators use their own covariance matrix corrections to compute heteroskedasticity and autocorrelation consistent (HAC), and cross-sectional or spatial correlation consistent standard errors under different data-generating assumptions, respectively. The various standard errors are given in parentheses. Significance levels: * p < 0.10, ** p < 0.05, *** p < 0.01.
|
Dependent variable: pse/gov
|
LSDV
|
LSDV
(One-way; Country)
|
LSDV
(One-way; Year)
|
LSDV
(Two-way)
|
Newey-West
|
Driscoll-Kraay
|
|
(A)
|
(B)
|
(A)
|
(B)
|
(A)
|
(B)
|
(A)
|
(B)
|
(A)
|
(B)
|
(A)
|
(B)
|
|
1#0. Low income, not democratic
|
3.633***
|
-0.012
|
3.633***
|
-0.012
|
3.633***
|
-0.012
|
3.633***
|
-0.012
|
3.633***
|
-0.012
|
3.633***
|
-0.012
|
|
(0.436)
|
(0.726)
|
(1.208)
|
(2.003)
|
(0.292)
|
(0.682)
|
(1.163)
|
(1.988)
|
(0.644)
|
(1.069)
|
(0.314)
|
(0.984)
|
|
1#1. Low income, democratic
|
4.796***
|
1.339*
|
4.796***
|
1.339
|
4.796***
|
1.339*
|
4.796***
|
1.339
|
4.796***
|
1.339
|
4.796***
|
1.339
|
|
(0.381)
|
(0.686)
|
(0.832)
|
(1.748)
|
(0.310)
|
(0.754)
|
(0.802)
|
(1.776)
|
(0.500)
|
(0.966)
|
(0.451)
|
(1.092)
|
|
2#0. Lower middle income, not democratic
|
4.304***
|
1.081*
|
4.304***
|
1.081
|
4.304***
|
1.081*
|
4.304***
|
1.081
|
4.304***
|
1.081
|
4.304***
|
1.081
|
|
(0.368)
|
(0.600)
|
(1.148)
|
(1.791)
|
(0.348)
|
(0.563)
|
(1.142)
|
(1.779)
|
(0.555)
|
(0.899)
|
(0.468)
|
(0.803)
|
|
2#1. Lower middle income, democratic
|
4.852***
|
2.421***
|
4.852***
|
2.421*
|
4.852***
|
2.421***
|
4.852***
|
2.421**
|
4.852***
|
2.421***
|
4.852***
|
2.421***
|
|
(0.369)
|
(0.446)
|
(1.009)
|
(1.236)
|
(0.214)
|
(0.390)
|
(0.963)
|
(1.217)
|
(0.548)
|
(0.639)
|
(0.219)
|
(0.440)
|
|
3#0. Upper middle income, not democratic
|
2.520***
|
-0.022
|
2.520*
|
-0.022
|
2.520***
|
-0.022
|
2.520*
|
-0.022
|
2.520***
|
-0.022
|
2.520***
|
-0.022
|
|
(0.455)
|
(0.503)
|
(1.380)
|
(1.502)
|
(0.461)
|
(0.378)
|
(1.382)
|
(1.465)
|
(0.663)
|
(0.749)
|
(0.668)
|
(0.547)
|
|
3#1. Upper middle income, democratic
|
3.643***
|
1.708***
|
3.643***
|
1.708
|
3.643***
|
1.708***
|
3.643***
|
1.708
|
3.643***
|
1.708***
|
3.643***
|
1.708**
|
|
(0.279)
|
(0.354)
|
(0.862)
|
(1.053)
|
(0.230)
|
(0.415)
|
(0.847)
|
(1.076)
|
(0.409)
|
(0.524)
|
(0.280)
|
(0.617)
|
|
4#0. High income (non-OECD), not democratic
|
0.987**
|
-2.057***
|
0.987
|
-2.057
|
0.987**
|
-2.057***
|
0.987
|
-2.057
|
0.987
|
-2.057**
|
0.987*
|
-2.057***
|
|
(0.499)
|
(0.566)
|
(1.699)
|
(1.648)
|
(0.378)
|
(0.407)
|
(1.668)
|
(1.600)
|
(0.770)
|
(0.830)
|
(0.516)
|
(0.339)
|
|
4#1. High income (non-OECD), democratic
|
0.568**
|
-0.038
|
0.568
|
-0.038
|
0.568***
|
-0.038
|
0.568
|
-0.038
|
0.568
|
-0.038
|
0.568**
|
-0.038
|
|
(0.249)
|
(0.254)
|
(0.834)
|
(0.815)
|
(0.188)
|
(0.211)
|
(0.818)
|
(0.803)
|
(0.372)
|
(0.379)
|
(0.260)
|
(0.284)
|
|
5#0. High income (OECD), not democratic
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
|
5#1. High income (OECD), democratic BASE
|
12.245***
|
4.332***
|
12.245***
|
4.332**
|
12.245***
|
4.332***
|
12.245***
|
4.332**
|
12.245***
|
4.332***
|
12.245***
|
4.332***
|
|
(0.122)
|
(0.831)
|
(0.500)
|
(2.156)
|
(0.088)
|
(0.602)
|
(0.493)
|
(2.079)
|
(0.196)
|
(1.172)
|
(0.128)
|
(0.621)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
pop024
|
|
0.146***
|
|
0.146***
|
|
0.146***
|
|
0.146***
|
|
0.146***
|
|
0.146***
|
|
|
(0.015)
|
|
(0.038)
|
|
(0.013)
|
|
(0.038)
|
|
(0.021)
|
|
(0.017)
|
|
urban
|
|
0.029***
|
|
0.029
|
|
0.029***
|
|
0.029
|
|
0.029***
|
|
0.029***
|
|
|
(0.007)
|
|
(0.020)
|
|
(0.006)
|
|
(0.020)
|
|
(0.010)
|
|
(0.009)
|
|
trade
|
|
0.017***
|
|
0.017***
|
|
0.017***
|
|
0.017***
|
|
0.017***
|
|
0.017***
|
|
|
(0.003)
|
|
(0.006)
|
|
(0.003)
|
|
(0.006)
|
|
(0.004)
|
|
(0.002)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
R-squared
|
0.141
|
0.198
|
0.141
|
0.198
|
0.141
|
0.198
|
0.141
|
0.198
|
0.141
|
0.198
|
0.141
|
0.198
|
|
F-value
|
64.55***
|
56.40***
|
7.16***
|
8.08***
|
234.33***
|
201.25***
|
n/a
|
n/a
|
29.30***
|
26.94***
|
315.69***
|
476.80***
|
|
Countries
|
175
|
165
|
175
|
165
|
175
|
165
|
175
|
165
|
175
|
165
|
175
|
165
|
|
Years
|
27
|
26
|
27
|
26
|
27
|
26
|
27
|
26
|
27
|
26
|
27
|
26
|
|
Observations
|
2194
|
2069
|
2194
|
2069
|
2194
|
2069
|
2194
|
2069
|
2194
|
2069
|
2194
|
2069
|
Notes: BASE group is high-income (OECD) and democratic countries. The model uses three controls: youth population (pop024); urban population (urban) and trade (trade). See the notes for table A4. The various standard errors are given in parentheses. Significance levels: * p < 0.10, ** p < 0.05, *** p < 0.01.
|
Dependent variable: pse/gdp
|
LSDV
|
LSDV
(One-way; Country)
|
LSDV
(One-way; Year)
|
LSDV
(Two-way)
|
Newey-West
|
Driscoll-Kraay
|
|
(A)
|
(B)
|
(A)
|
(B)
|
(A)
|
(B)
|
(A)
|
(B)
|
(A)
|
(B)
|
(A)
|
(B)
|
|
0#0. Poorer country regions, not democratic
|
-1.298***
|
-2.031***
|
-1.298***
|
-2.031***
|
-1.298***
|
-2.031***
|
-1.298***
|
-2.031***
|
-1.298***
|
-2.031***
|
-1.298***
|
-2.031***
|
|
(0.147)
|
(0.216)
|
(0.396)
|
(0.582)
|
(0.128)
|
(0.168)
|
(0.389)
|
(0.566)
|
(0.199)
|
(0.315)
|
(0.152)
|
(0.218)
|
|
0#1. Poorer country regions, democratic
|
-0.804***
|
-1.412***
|
-0.804**
|
-1.412***
|
-0.804***
|
-1.412***
|
-0.804**
|
-1.412***
|
-0.804***
|
-1.412***
|
-0.804***
|
-1.412***
|
|
(0.104)
|
(0.166)
|
(0.343)
|
(0.467)
|
(0.081)
|
(0.135)
|
(0.336)
|
(0.457)
|
(0.158)
|
(0.239)
|
(0.110)
|
(0.162)
|
|
1#0. Richer country regions, not democratic
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
|
1#1. Richer country regions, democratic BASE
|
5.169***
|
2.095***
|
5.169***
|
2.095**
|
5.169***
|
2.095***
|
5.169***
|
2.095**
|
5.169***
|
2.095***
|
5.169***
|
2.095***
|
|
(0.069)
|
(0.375)
|
(0.264)
|
(0.989)
|
(0.052)
|
(0.255)
|
(0.260)
|
(0.950)
|
(0.110)
|
(0.534)
|
(0.064)
|
(0.319)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
pop024
|
|
0.034***
|
|
0.034**
|
|
0.034***
|
|
0.034**
|
|
0.034***
|
|
0.034***
|
|
|
(0.007)
|
|
(0.017)
|
|
(0.005)
|
|
(0.016)
|
|
(0.010)
|
|
(0.005)
|
|
urban
|
|
0.016***
|
|
0.016**
|
|
0.016***
|
|
0.016**
|
|
0.016***
|
|
0.016***
|
|
|
(0.003)
|
|
(0.008)
|
|
(0.002)
|
|
(0.008)
|
|
(0.004)
|
|
(0.002)
|
|
trade
|
|
0.013***
|
|
0.013***
|
|
0.013***
|
|
0.013***
|
|
0.013***
|
|
0.013***
|
|
|
(0.002)
|
|
(0.004)
|
|
(0.002)
|
|
(0.004)
|
|
(0.002)
|
|
(0.002)
|
|
R-squared
|
0.057
|
0.167
|
0.057
|
0.167
|
0.057
|
0.167
|
0.057
|
0.167
|
0.057
|
0.167
|
0.057
|
0.167
|
|
F-value
|
52.61***
|
63.60***
|
5.45***
|
6.80***
|
148.23***
|
109.08***
|
n/a
|
n/a
|
24.89***
|
31.33***
|
161.84***
|
103.56***
|
|
Countries
|
102
|
97
|
102
|
97
|
102
|
97
|
102
|
97
|
102
|
97
|
102
|
97
|
|
Years
|
27
|
26
|
27
|
26
|
27
|
26
|
27
|
26
|
27
|
26
|
27
|
26
|
|
Observations
|
1486
|
1382
|
1486
|
1382
|
1486
|
1382
|
1486
|
1382
|
1486
|
1382
|
1486
|
1382
|
Notes: BASE group is richer country regions that are democratic. The model uses three controls: youth population (pop024); urban population (urban) and trade (trade). See the notes for table A4. The various standard errors are given in parentheses. Significance levels are as follows: * p < 0.10, ** p < 0.05, *** p < 0.01.
|
Dependent variable: pse/gov
|
LSDV
|
LSDV
(One-way; Country)
|
LSDV
(One-way; Year)
|
LSDV
(Two-way)
|
Newey-West
|
Driscoll-Kraay
|
|
(A)
|
(B)
|
(A)
|
(B)
|
(A)
|
(B)
|
(A)
|
(B)
|
(A)
|
(B)
|
(A)
|
(B)
|
|
0#0. Poorer country regions, not democratic
|
3.880***
|
1.492**
|
3.880***
|
1.492
|
3.880***
|
1.492***
|
3.880***
|
1.492
|
3.880***
|
1.492*
|
3.880***
|
1.492***
|
|
(0.305)
|
(0.612)
|
(0.909)
|
(1.609)
|
(0.229)
|
(0.360)
|
(0.887)
|
(1.531)
|
(0.452)
|
(0.892)
|
(0.210)
|
(0.466)
|
|
0#1. Poorer country regions, democratic
|
5.077***
|
3.321***
|
5.077***
|
3.321**
|
5.077***
|
3.321***
|
5.077***
|
3.321**
|
5.077***
|
3.321***
|
5.077***
|
3.321***
|
|
(0.233)
|
(0.458)
|
(0.711)
|
(1.397)
|
(0.141)
|
(0.236)
|
(0.686)
|
(1.341)
|
(0.345)
|
(0.684)
|
(0.157)
|
(0.301)
|
|
1#0. Richer country regions, not democratic
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
|
1#1. Richer country regions, democratic BASE
|
11.944***
|
5.333***
|
11.944***
|
5.333
|
11.944***
|
5.333***
|
11.944***
|
5.333*
|
11.944***
|
5.333***
|
11.944***
|
5.333***
|
|
(0.127)
|
(1.193)
|
(0.506)
|
(3.265)
|
(0.116)
|
(0.952)
|
(0.504)
|
(3.185)
|
(0.203)
|
(1.719)
|
(0.176)
|
(1.039)
|
|
pop024
|
|
0.104***
|
|
0.104*
|
|
0.104***
|
|
0.104*
|
|
0.104***
|
|
0.104***
|
|
|
(0.021)
|
|
(0.059)
|
|
(0.013)
|
|
(0.056)
|
|
(0.030)
|
|
(0.015)
|
|
urban
|
|
0.029***
|
|
0.029
|
|
0.029***
|
|
0.029
|
|
0.029**
|
|
0.029***
|
|
|
(0.008)
|
|
(0.023)
|
|
(0.008)
|
|
(0.023)
|
|
(0.012)
|
|
(0.009)
|
|
trade
|
|
0.018***
|
|
0.018**
|
|
0.018***
|
|
0.018**
|
|
0.018***
|
|
0.018***
|
|
|
(0.004)
|
|
(0.009)
|
|
(0.004)
|
|
(0.009)
|
|
(0.005)
|
|
(0.004)
|
|
R-squared
|
0.195
|
0.222
|
0.195
|
0.222
|
0.195
|
0.222
|
0.195
|
0.222
|
0.195
|
0.222
|
0.195
|
0.222
|
|
F-value
|
270.00***
|
113.21***
|
25.86***
|
13.06***
|
723.40***
|
323.35***
|
n/a
|
n/a
|
118.36***
|
51.83***
|
611.34***
|
290.39***
|
|
Countries
|
99
|
96
|
99
|
96
|
99
|
96
|
99
|
96
|
99
|
96
|
99
|
96
|
|
Years
|
27
|
26
|
27
|
26
|
27
|
26
|
27
|
26
|
27
|
26
|
27
|
26
|
|
Observations
|
1360
|
1299
|
1360
|
1299
|
1360
|
1299
|
1360
|
1299
|
1360
|
1299
|
1360
|
1299
|
Notes: BASE group is richer country regions that are democratic. The model uses three controls: youth population (pop024); urban population (urban) and trade (trade). See the notes for table A4. The various standard errors are given in parentheses. Significance levels: * p < 0.10, ** p < 0.05, *** p < 0.01.
|
Dependent variable: pse/gdp
|
I
|
II
|
III
|
IV
|
V
|
VI
|
VII
|
VIII
|
IX
|
X
|
XI
|
XII
|
|
0#0. Poorer country regions, not democratic
|
-2.031***
|
-2.183***
|
-1.764***
|
-1.866***
|
-0.991***
|
-1.072***
|
-0.568*
|
-0.692*
|
-0.800***
|
-1.146***
|
-0.906***
|
-1.163***
|
|
(0.216)
|
(0.242)
|
(0.224)
|
(0.258)
|
(0.275)
|
(0.365)
|
(0.296)
|
(0.383)
|
(0.268)
|
(0.293)
|
(0.279)
|
(0.307)
|
|
0#1. Poorer country regions, democratic
|
-1.412***
|
-1.543***
|
-1.379***
|
-1.448***
|
-0.578**
|
-0.631**
|
-0.028
|
-0.102
|
-0.021
|
-0.277
|
-0.042
|
-0.241
|
|
(0.166)
|
(0.155)
|
(0.176)
|
(0.167)
|
(0.237)
|
(0.288)
|
(0.258)
|
(0.315)
|
(0.257)
|
(0.279)
|
(0.265)
|
(0.288)
|
|
1#0. Richer country regions, not democratic
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
|
1#1. Richer country regions, democratic BASE
|
2.095***
|
1.504***
|
-1.863**
|
-2.243***
|
-5.403***
|
-5.626***
|
-7.503***
|
-7.711***
|
-6.685***
|
-6.649***
|
-6.539***
|
-6.743***
|
|
(0.375)
|
(0.518)
|
(0.760)
|
(0.790)
|
(1.427)
|
(1.549)
|
(1.550)
|
(1.713)
|
(1.241)
|
(1.295)
|
(1.297)
|
(1.370)
|
|
R-squared
|
0.167
|
0.187
|
0.201
|
0.223
|
0.208
|
0.229
|
0.223
|
0.245
|
0.353
|
0.375
|
0.365
|
0.384
|
|
F-value
|
63.60***
|
15.33***
|
103.09***
|
23.20***
|
91.08***
|
22.81***
|
80.33***
|
21.89***
|
69.88***
|
21.30***
|
60.49***
|
19.72***
|
|
Countries
|
97
|
97
|
86
|
86
|
86
|
86
|
85
|
85
|
84
|
84
|
84
|
84
|
|
Years
|
26
|
26
|
25
|
25
|
25
|
25
|
25
|
25
|
25
|
25
|
25
|
25
|
|
Year fixed effects
|
No
|
Yes
|
No
|
Yes
|
No
|
Yes
|
No
|
Yes
|
No
|
Yes
|
No
|
Yes
|
|
Covariates (controls)
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
|
Number of control variables
|
3
|
3
|
4
|
4
|
5
|
5
|
6
|
6
|
7
|
7
|
8
|
8
|
|
Are patterns confounded?
|
No
|
No
|
No
|
No
|
No
|
No
|
Partially
|
Partially
|
Partially
|
Partially
|
Partially
|
Partially
|
|
Observations
|
1382
|
1382
|
1256
|
1256
|
1256
|
1256
|
1174
|
1174
|
1101
|
1101
|
1030
|
1030
|
Notes: BASE group is richer country regions that are democratic. All models use a LSDV estimator and robust standard errors. I and II use homogeneous slopes and three controls (pop024, urban and trade). III and IV use homogeneous slopes and four controls (pop024, urban, trade and hci). V and VI use homogeneous slopes and five controls (pop024, urban, trade, hci and pop65). VII and VIII use homogeneous slopes and six controls (pop024, urban, trade, hci, pop65 and military). IX and X use homogeneous slopes and seven controls (pop024, urban, trade, hci, pop65, military and fiscbal). XI and XII use homogeneous slopes and eight controls (pop024, urban, trade, hci, pop65, military, fiscbal and debt). See table 1 for a description of each control variable used. Time (year) dummies are used in even-numbered specifications (II, IV, VI, VIII, X and XII). The estimates for the various controls and year fixed effects are excluded to save space. Huber-White heteroskedasticity-robust standard errors are given in parentheses. Significance levels: * p < 0.10, ** p < 0.05, *** p < 0.01.
|
Dependent variable: pse/gov
|
I
|
II
|
III
|
IV
|
V
|
VI
|
VII
|
VIII
|
IX
|
X
|
XI
|
XII
|
|
0#0. Poorer country regions, not democratic
|
1.492**
|
0.985
|
1.296**
|
0.737
|
1.358**
|
0.499
|
2.207***
|
1.514**
|
1.962***
|
1.269*
|
1.803***
|
1.668**
|
|
(0.612)
|
(0.664)
|
(0.644)
|
(0.700)
|
(0.637)
|
(0.718)
|
(0.661)
|
(0.755)
|
(0.656)
|
(0.749)
|
(0.667)
|
(0.756)
|
|
0#1. Poorer country regions, democratic
|
3.321***
|
2.799***
|
2.970***
|
2.451***
|
3.035***
|
2.204***
|
3.784***
|
3.186***
|
3.717***
|
3.162***
|
3.561***
|
3.391***
|
|
(0.458)
|
(0.499)
|
(0.481)
|
(0.530)
|
(0.518)
|
(0.596)
|
(0.548)
|
(0.628)
|
(0.539)
|
(0.621)
|
(0.552)
|
(0.630)
|
|
1#0. Richer country regions, not democratic
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
|
1#1. Richer country regions, democratic BASE
|
5.333***
|
3.726***
|
-3.448**
|
-4.803***
|
-3.737
|
-3.794
|
-4.396
|
-4.194
|
-1.745
|
-1.109
|
-1.285
|
-1.221
|
|
(1.193)
|
(1.443)
|
(1.686)
|
(1.810)
|
(2.738)
|
(2.805)
|
(2.840)
|
(2.923)
|
(2.867)
|
(2.948)
|
(2.897)
|
(2.955)
|
|
R-squared
|
0.222
|
0.237
|
0.296
|
0.311
|
0.296
|
0.311
|
0.331
|
0.343
|
0.347
|
0.360
|
0.415
|
0.423
|
|
F-value
|
113.21***
|
20.02***
|
110.62***
|
23.58***
|
100.86***
|
24.23***
|
91.63***
|
23.86***
|
92.91***
|
27.10***
|
96.81***
|
31.64***
|
|
Countries
|
96
|
96
|
85
|
85
|
85
|
85
|
84
|
84
|
84
|
84
|
84
|
84
|
|
Years
|
26
|
26
|
25
|
25
|
25
|
25
|
25
|
25
|
25
|
25
|
25
|
25
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Year fixed effects
|
No
|
Yes
|
No
|
Yes
|
No
|
Yes
|
No
|
Yes
|
No
|
Yes
|
No
|
Yes
|
|
Covariates (controls)
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
|
Number of control variables
|
3
|
3
|
4
|
4
|
5
|
5
|
6
|
6
|
7
|
7
|
8
|
8
|
|
Are patterns confounded?
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
|
Observations
|
1299
|
1299
|
1175
|
1175
|
1175
|
1175
|
1099
|
1099
|
1091
|
1091
|
1024
|
1024
|
Notes: BASE group is richer country regions that are democratic. All models use a LSDV estimator and robust standard errors. See the notes for table A8. Huber-White heteroskedasticity-robust standard errors are given in parentheses. Significance levels are as follows: * p < 0.10, ** p < 0.05, *** p < 0.01.
|
Dependent variable: pse/gdp
|
I
|
II
|
III
|
IV
|
V
|
VI
|
VII
|
VIII
|
IX
|
X
|
XI
|
XII
|
|
0#0. Poorer country regions, not democratic
|
-1.760***
|
-2.022***
|
-1.489***
|
-1.729***
|
-1.240***
|
-1.615***
|
-0.926***
|
-1.311***
|
-0.930***
|
-1.332***
|
-1.077***
|
-1.418***
|
|
(0.199)
|
(0.206)
|
(0.199)
|
(0.208)
|
(0.260)
|
(0.275)
|
(0.274)
|
(0.291)
|
(0.279)
|
(0.297)
|
(0.274)
|
(0.297)
|
|
0#1. Poorer country regions, democratic
|
-1.178***
|
-1.403***
|
-1.111***
|
-1.310***
|
-0.844***
|
-1.188***
|
-0.455*
|
-0.785***
|
-0.424*
|
-0.734***
|
-0.522**
|
-0.797***
|
|
(0.166)
|
(0.172)
|
(0.165)
|
(0.172)
|
(0.239)
|
(0.252)
|
(0.253)
|
(0.267)
|
(0.256)
|
(0.271)
|
(0.250)
|
(0.268)
|
|
1#0. Richer country regions, not democratic
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
|
1#1. Richer country regions, democratic BASE
|
2.520***
|
1.764***
|
0.251
|
-0.321
|
-1.013
|
-0.848
|
-2.635***
|
-2.416**
|
-3.139***
|
-2.862***
|
-2.688***
|
-2.624**
|
|
(0.348)
|
(0.425)
|
(0.520)
|
(0.566)
|
(0.924)
|
(0.951)
|
(0.965)
|
(0.998)
|
(1.002)
|
(1.041)
|
(0.988)
|
(1.053)
|
|
R-squared
|
0.313
|
0.335
|
0.331
|
0.351
|
0.333
|
0.351
|
0.338
|
0.357
|
0.340
|
0.363
|
0.373
|
0.392
|
|
F-value
|
125.25***
|
22.67***
|
103.17***
|
22.06***
|
88.81***
|
21.32***
|
74.25***
|
19.81***
|
62.34***
|
18.39***
|
60.71***
|
18.87***
|
|
Countries
|
97
|
97
|
86
|
86
|
86
|
86
|
85
|
85
|
84
|
84
|
84
|
84
|
|
Years
|
26
|
26
|
25
|
25
|
25
|
25
|
25
|
25
|
25
|
25
|
25
|
25
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Year fixed effects
|
No
|
Yes
|
No
|
Yes
|
No
|
Yes
|
No
|
Yes
|
No
|
Yes
|
No
|
Yes
|
|
Covariates (controls)
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
|
Number of control variables
|
3
|
3
|
4
|
4
|
5
|
5
|
6
|
6
|
7
|
7
|
8
|
8
|
|
Are the patterns confounded?
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
|
Observations
|
1382
|
1382
|
1256
|
1256
|
1256
|
1256
|
1174
|
1174
|
1101
|
1101
|
1030
|
1030
|
Notes: BASE group is richer country regions that are democratic. All models use a robust estimator. See the notes for table A8. Significance levels: * p < 0.10, ** p < 0.05, *** p < 0.01.
|
Dependent variable: pse/gov
|
I
|
II
|
III
|
IV
|
V
|
VI
|
VII
|
VIII
|
IX
|
X
|
XI
|
XII
|
|
0#0. Poorer country regions, not democratic
|
1.108*
|
0.522
|
1.000*
|
0.310
|
1.122
|
0.172
|
1.972**
|
1.255
|
1.798**
|
1.077
|
1.614**
|
1.315*
|
|
(0.597)
|
(0.626)
|
(0.588)
|
(0.617)
|
(0.761)
|
(0.806)
|
(0.771)
|
(0.829)
|
(0.751)
|
(0.804)
|
(0.702)
|
(0.767)
|
|
0#1. Poorer country regions, democratic
|
2.945***
|
2.335***
|
2.626***
|
1.919***
|
2.760***
|
1.777**
|
3.292***
|
2.617***
|
3.268***
|
2.608***
|
3.037***
|
2.724***
|
|
(0.491)
|
(0.517)
|
(0.479)
|
(0.503)
|
(0.697)
|
(0.737)
|
(0.709)
|
(0.757)
|
(0.690)
|
(0.734)
|
(0.642)
|
(0.694)
|
|
1#0. Richer country regions, not democratic
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
|
1#1. Richer country regions, democratic BASE
|
3.705***
|
1.652
|
-4.570***
|
-6.683***
|
-5.296*
|
-5.963**
|
-5.074*
|
-5.589**
|
-2.121
|
-2.119
|
-1.306
|
-1.495
|
|
(1.036)
|
(1.400)
|
(1.533)
|
(1.748)
|
(2.708)
|
(2.810)
|
(2.701)
|
(2.838)
|
(2.692)
|
(2.821)
|
(2.531)
|
(2.719)
|
|
R-squared
|
0.247
|
0.264
|
0.328
|
0.348
|
0.328
|
0.347
|
0.373
|
0.386
|
0.402
|
0.421
|
0.483
|
0.492
|
|
F-value
|
84.82***
|
15.18***
|
94.92***
|
20.35***
|
81.47***
|
19.61***
|
81.11***
|
20.98***
|
80.84***
|
23.30***
|
94.69***
|
28.19***
|
|
Countries
|
96
|
96
|
85
|
85
|
85
|
85
|
84
|
84
|
84
|
84
|
84
|
84
|
|
Years
|
26
|
26
|
25
|
25
|
25
|
25
|
25
|
25
|
25
|
25
|
25
|
25
|
|
Year fixed effects
|
No
|
Yes
|
No
|
Yes
|
No
|
Yes
|
No
|
Yes
|
No
|
Yes
|
No
|
Yes
|
|
Covariates (controls)
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
Yes
|
|
Number of control variables
|
3
|
3
|
4
|
4
|
5
|
5
|
6
|
6
|
7
|
7
|
8
|
8
|
|
Are the patterns confounded?
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
|
Observations
|
1299
|
1299
|
1175
|
1175
|
1175
|
1175
|
1099
|
1099
|
1091
|
1091
|
1024
|
1024
|
Notes: BASE group is richer country regions that are democratic. All models use a robust estimator. See the notes for table A8. Significance levels: * p < 0.10, ** p < 0.05, *** p < 0.01.
|
Dependent variable: pse/gdp
|
I
|
II
|
III
|
IV
|
V
|
VI
|
VII
|
VIII
|
IX
|
X
|
|
1#0. Low income, not democratic
|
|
-0.664
|
-1.190***
|
-2.095***
|
-2.081***
|
-1.961***
|
-1.885***
|
-1.810***
|
-2.120***
|
-2.366***
|
|
|
(0.408)
|
(0.224)
|
(0.185)
|
(0.188)
|
(0.261)
|
(0.260)
|
(0.243)
|
(0.301)
|
(0.350)
|
|
1#1. Low income, democratic
|
|
-0.367
|
-1.244***
|
-1.622***
|
-1.608***
|
-1.846***
|
-1.877***
|
-1.851***
|
-2.070***
|
-2.220***
|
|
|
(0.434)
|
(0.228)
|
(0.184)
|
(0.202)
|
(0.195)
|
(0.257)
|
(0.252)
|
(0.324)
|
(0.312)
|
|
2#0. Lower middle income, not democratic
|
|
-0.784***
|
-1.426***
|
-2.093***
|
-2.050***
|
-1.845***
|
-1.578***
|
-1.210***
|
-1.128***
|
-1.259***
|
|
|
(0.275)
|
(0.178)
|
(0.168)
|
(0.155)
|
(0.206)
|
(0.259)
|
(0.237)
|
(0.274)
|
(0.267)
|
|
2#1. Lower middle income, democratic
|
|
-0.156
|
-1.150***
|
-1.644***
|
-1.628***
|
-1.491***
|
-1.126***
|
-0.530**
|
-0.144
|
-0.193
|
|
|
(0.227)
|
(0.161)
|
(0.143)
|
(0.133)
|
(0.211)
|
(0.287)
|
(0.267)
|
(0.289)
|
(0.264)
|
|
3#0. Upper middle income, not democratic
|
|
-1.256***
|
-1.329***
|
-1.927***
|
-1.857***
|
-1.744***
|
-1.669***
|
-1.604***
|
-1.729***
|
-1.803***
|
|
|
(0.224)
|
(0.171)
|
(0.125)
|
(0.157)
|
(0.125)
|
(0.177)
|
(0.152)
|
(0.202)
|
(0.224)
|
|
3#1. Upper middle income, democratic
|
|
-0.910***
|
-0.983***
|
-1.192***
|
-1.182***
|
-1.205***
|
-1.131***
|
-1.017***
|
-1.175***
|
-1.344***
|
|
|
(0.185)
|
(0.128)
|
(0.110)
|
(0.105)
|
(0.105)
|
(0.156)
|
(0.127)
|
(0.141)
|
(0.178)
|
|
4#0. High income (non-OECD), not democratic
|
|
n/a
|
-1.619***
|
-2.149***
|
-1.974***
|
-2.025***
|
-1.933***
|
-2.164***
|
-1.911***
|
-1.959***
|
|
|
n/a
|
(0.205)
|
(0.258)
|
(0.323)
|
(0.135)
|
(0.195)
|
(0.222)
|
(0.424)
|
(0.377)
|
|
4#1. High income (non-OECD), democratic
|
|
-0.565***
|
-0.645***
|
-0.801***
|
-0.748***
|
-0.667***
|
-0.654***
|
-0.531***
|
-0.578***
|
-0.654***
|
|
|
(0.156)
|
(0.107)
|
(0.100)
|
(0.090)
|
(0.121)
|
(0.130)
|
(0.115)
|
(0.150)
|
(0.149)
|
|
5#0. High income (OECD), not democratic
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
|
5#1. High income (OECD), democratic BASE
|
|
3.203***
|
1.755***
|
3.722***
|
3.603***
|
3.047***
|
3.107***
|
3.000***
|
2.708***
|
2.452***
|
|
|
(0.450)
|
(0.417)
|
(0.224)
|
(0.223)
|
(0.277)
|
(0.334)
|
(0.301)
|
(0.381)
|
(0.391)
|
|
Constant
|
2.963***
|
|
|
|
|
|
|
|
|
|
|
(0.287)
|
|
|
|
|
|
|
|
|
|
|
gdppc
|
0.000009***
|
|
|
|
|
|
|
|
|
|
|
(0.000003)
|
|
|
|
|
|
|
|
|
|
|
poldemoc
|
0.713***
|
|
|
|
|
|
|
|
|
|
|
(0.076)
|
|
|
|
|
|
|
|
|
|
|
gini
|
|
-0.005
|
|
|
|
|
|
|
|
|
|
|
(0.008)
|
|
|
|
|
|
|
|
|
|
hci (lagged
one period)
|
|
|
0.499***
|
|
|
|
|
|
|
|
|
|
|
(0.091)
|
|
|
|
|
|
|
|
|
R-squared
|
0.158
|
0.280
|
0.213
|
0.151
|
0.144
|
0.120
|
0.107
|
0.095
|
0.084
|
0.089
|
|
F-value
|
85.07***
|
26.72***
|
45.06***
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
|
Countries
|
167
|
124
|
137
|
169
|
169
|
169
|
169
|
169
|
169
|
169
|
|
Years
|
26
|
25
|
26
|
26
|
26
|
26
|
26
|
26
|
26
|
26
|
|
Year fixed effects
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
|
Observations
|
2278
|
769
|
2010
|
2288
|
2288
|
2288
|
2288
|
2288
|
2288
|
2288
|
Notes: BASE group is high-income (OECD) and democratic countries. Models I, II, and III use a robust estimator. Models IV, V, VI, VII, VIII, IX, and X use a quantile estimator at the 20th, 25th, 40th, 50th, 60th, 75th, and 80th percentiles, respectively. The constant for Model I includes all countries, regardless of income group, ceteris paribus. All models use three controls: youth population (pop024); urban population (urban) and trade (trade). The estimates for the various controls are excluded to save space. A pseudo R-squared is reported for the quantile regression. Not applicable (n/a) means there are no applicable countries for these groups. The various standard errors are given in parentheses. Significance levels: * p < 0.10, ** p < 0.05, *** p < 0.01.
|
Dependent variable: pse/gov
|
I
|
II
|
III
|
IV
|
V
|
VI
|
VII
|
VIII
|
IX
|
X
|
|
1#0. Low income, not democratic
|
|
0.967
|
1.180*
|
-0.341
|
-0.274
|
0.052
|
-0.162
|
-1.109*
|
-0.854
|
-0.754
|
|
|
(1.063)
|
(0.656)
|
(0.787)
|
(0.700)
|
(0.972)
|
(0.696)
|
(0.673)
|
(0.991)
|
(1.007)
|
|
1#1. Low income, democratic
|
|
1.933*
|
0.999
|
3.255***
|
2.717***
|
2.435***
|
1.290**
|
-0.190
|
-0.770
|
-0.523
|
|
|
(1.137)
|
(0.666)
|
(0.765)
|
(0.539)
|
(0.745)
|
(0.581)
|
(0.475)
|
(1.060)
|
(1.045)
|
|
2#0. Lower middle income, not democratic
|
|
0.891
|
-0.048
|
0.215
|
0.286
|
0.918
|
0.614
|
0.335
|
0.879
|
0.581
|
|
|
(0.729)
|
(0.519)
|
(0.531)
|
(0.568)
|
(0.676)
|
(0.531)
|
(0.492)
|
(0.771)
|
(0.753)
|
|
2#1. Lower middle income, democratic
|
|
1.694***
|
1.710***
|
2.115***
|
1.905***
|
1.878***
|
1.661***
|
1.506***
|
1.885**
|
1.730**
|
|
|
(0.588)
|
(0.465)
|
(0.286)
|
(0.120)
|
(0.547)
|
(0.419)
|
(0.253)
|
(0.743)
|
(0.814)
|
|
3#0. Upper middle income, not democratic
|
|
0.323
|
1.569***
|
-2.129**
|
-1.405*
|
0.003
|
-0.131
|
-0.353
|
1.465**
|
1.055
|
|
|
(0.584)
|
(0.506)
|
(0.860)
|
(0.802)
|
(0.638)
|
(0.371)
|
(0.509)
|
(0.717)
|
(0.755)
|
|
3#1. Upper middle income, democratic
|
|
0.237
|
1.188***
|
0.811***
|
0.831***
|
1.344***
|
1.205***
|
0.967***
|
1.514***
|
1.371**
|
|
|
(0.478)
|
(0.370)
|
(0.236)
|
(0.263)
|
(0.371)
|
(0.179)
|
(0.254)
|
(0.553)
|
(0.592)
|
|
4#0. High income (non-OECD), not democratic
|
|
n/a
|
-1.415**
|
-2.972***
|
-3.211***
|
-3.372***
|
-3.107***
|
-3.757***
|
-1.624
|
-1.034
|
|
|
n/a
|
(0.581)
|
(0.304)
|
(0.387)
|
(0.627)
|
(0.356)
|
(0.303)
|
(1.293)
|
(1.085)
|
|
4#1. High income (non-OECD), democratic
|
|
-0.117
|
-0.112
|
-0.914***
|
-0.685**
|
0.065
|
0.308
|
0.413
|
0.821**
|
0.326
|
|
|
(0.400)
|
(0.320)
|
(0.218)
|
(0.327)
|
(0.425)
|
(0.342)
|
(0.278)
|
(0.401)
|
(0.425)
|
|
5#0. High income (OECD), not democratic
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
|
5#1. High income (OECD), democratic BASE
|
|
1.036
|
0.998
|
4.287***
|
3.603***
|
2.246**
|
0.885
|
-0.810
|
1.296
|
3.350***
|
|
|
(1.164)
|
(1.225)
|
(0.481)
|
(0.664)
|
(0.907)
|
(0.643)
|
(0.726)
|
(1.110)
|
(1.064)
|
|
Constant
|
1.989**
|
|
|
|
|
|
|
|
|
|
|
(0.841)
|
|
|
|
|
|
|
|
|
|
|
gdppc
|
-0.000015**
|
|
|
|
|
|
|
|
|
|
|
(0.000007)
|
|
|
|
|
|
|
|
|
|
|
poldemoc
|
1.368***
|
|
|
|
|
|
|
|
|
|
|
(0.229)
|
|
|
|
|
|
|
|
|
|
|
gini
|
|
0.098***
|
|
|
|
|
|
|
|
|
|
|
(0.021)
|
|
|
|
|
|
|
|
|
|
hci (lagged
one period)
|
|
|
0.404
|
|
|
|
|
|
|
|
|
|
|
(0.271)
|
|
|
|
|
|
|
|
|
R-squared
|
0.205
|
0.295
|
0.279
|
0.081
|
0.093
|
0.121
|
0.146
|
0.166
|
0.189
|
0.193
|
|
F-value
|
105.86***
|
26.91***
|
58.31***
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
n/a
|
|
Countries
|
164
|
124
|
134
|
165
|
165
|
165
|
165
|
165
|
165
|
165
|
|
Years
|
26
|
24
|
26
|
26
|
26
|
26
|
26
|
26
|
26
|
26
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Year fixed effects
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
No
|
|
Observations
|
2060
|
718
|
1820
|
2069
|
2069
|
2069
|
2069
|
2069
|
2069
|
2069
|
Notes: BASE group is high-income (OECD) and democratic countries. Models I, II, and III use a robust estimator. Models IV, V, VI, VII, VIII, IX, and X use a quantile estimator at the 20th, 25th, 40th, 50th, 60th, 75th, and 80th percentiles, respectively. See the notes for table A12. Significance levels: * p < 0.10, ** p < 0.05, *** p < 0.01.
Table 1Data definitions and sources DISPLAY Table
Table 2Mean differences in the national effort and budget share by income group and regime type DISPLAY Table
Table 3Mean differences in the national effort and budget share by country region and regime type DISPLAY Table
Table 4Wald tests for parameter equality of the factor-variable interactions DISPLAY Table
Table 5Three inequality propositions DISPLAY Table
Table A1List of countries and territories by GNI per capita group in 2015 (ypc2015) DISPLAY Table
Table A2List of countries by two regional country groups (region) DISPLAY Table
Table A3Descriptive statistics DISPLAY Table
Table A4Mean differences in the national effort by income group and regime type – alternative standard error estimates DISPLAY Table
Table A5Mean differences in the budget share by income group and regime type – alternative standard error estimates DISPLAY Table
Table A6Mean differences in the national effort by country region and regime type – alternative standard error estimates DISPLAY Table
Table A7Mean differences in the budget share by country region and regime type – alternative standard error estimates DISPLAY Table
Table A8A summary of changes to the model specification (national effort and LSDV estimator) DISPLAY Table
Table A9A summary of changes to the model specification (budget share and LSDV estimator) DISPLAY Table
Table A10A summary of changes to the model specification (national effort and robust estimator) DISPLAY Table
Table A11A summary of changes to the model specification (budget share and robust estimator) DISPLAY Table
Table A12A summary of additional robustness checks (national effort with robust and quantile estimators) DISPLAY Table
Table A13A summary of additional robustness checks (budget share with robust and quantile estimators) DISPLAY Table
* We thank two anonymous reviewers, and participants at the 59th Annual Conference of the New Zealand Association of Economists and at the University of Otago, Department of Economics, Brown Bag seminar for helpful comments and suggestions.
1 More detailed explanations of the compensation and efficiency hypotheses are provided by Walter (  2010), Adserà and Boix ( 2002), Garrett ( 1998a, 1998b, 2001), Rodrik ( 1998), Katzenstein ( 1985), Ruggie ( 1982) and Cameron ( 1978).
2 Robust estimation uses the “rreg” routine in Stata. An initial screening based on Cook’s distance is used to remove gross outliers. Starting values are then calculated, and Huber iterations performed, followed by biweight iterations, to determine the down-weighting of any outliers; see Hamilton ( 1991) for further details.
3 Baum, Nichols and Schaffer ( 2010) and Cameron and Miller ( 2015) provide a practical discussion of clusterrobust inference. All estimates are obtained using Stata; one-way clustering of standard errors is performed using “cluster(country)” or “cluster(year)”. Two-way clustering is performed with the user-written program “vce2way” (Yoo, 2017).
4 We are grateful to a reviewer for suggesting these additional robustness checks.
5 Partial confounding refers to the case where only poorer countries that are not democratic are shown to have significantly different means from the base group (richer and democratic countries), and with the expected sign. No confounding refers to the case where either poorer country groups (irrespective of the state of democracy) or poorer and democratic countries are shown to have significantly different means from the base group, and with the expected sign.
6 Alternative specifications were also fitted using GDP per capita and its squared and cubed values, along with the political democracy variable and main controls. In all cases considered, the main results are supported, i.e., the coefficient on the linear GDP per capita term maintains the same sign, is not too dissimilar in size, and remains statistically significant. Note that, in the main results, non-linearities are allowed for by estimating piecewise linear effects, i.e., average effects for different income groupings of countries.
7 We note two points relating to these inequalities. First, it does not matter whether E, Y and G are measured in real or nominal terms, provided both the numerator and denominator of the relevant ratio are measured in the same nominal or real terms (using the same deflator). Second, the same estimated size of the public sector in any one country, as given by sources such as the IMF, cannot simply be obtained by taking the quotient of the national effort and budget share for that country because these education spending measures are estimates. The quotient will give only a rough approximation of the size of government, especially for countries that have less accurate education spending data.
|
|
September, 2023
III/2023
|
