16182
Views
934
Downloads |
Decentralization and welfare: theory and an empirical analysis using Philippine data
Tristan Canare* 
Tristan Canare
Affiliation: Department of Economics, School of Social Sciences, Ateneo de Manila University, Katipunan Ave., Loyola Heights, Quezon City, Metro Manila, Philippines 1108
 0000-0001-7737-3217
Correspondence
tristan.canare@gmail.com
Article | Year: 2021 | Pages: 93 - 123 | Volume: 45 | Issue: 1 Received: July 9, 2020 | Accepted: November 28, 2020 | Published online: March 9, 2021
|
FULL ARTICLE
FIGURES & DATA
REFERENCES
CROSSMARK POLICY
METRICS
LICENCING
PDF
|
Government Setup
|
Utility Function
|
|
Constraint
|
|
Decentralized
|
Ui = Xi + θigβ
|
(1)
|
|
M = X + P * g
|
(3)
|
|
Centralized
|
Ui=Xi + (1-ω) θigβ
|
(2)
|
|
M = X + (1-σ)P*g
|
(4)
|
|
|
Income per capita
|
HDI
|
Poverty
|
|
(1)
|
(2)
|
(3)
|
(4)
|
(5)
|
(6)
|
(7)
|
(8)
|
(9)
|
|
|
|
|
|
|
|
|
|
|
|
|
fiscal_indep
|
10,935***
|
14,638***
|
8,981*
|
0.0701**
|
0.0847***
|
0.0776**
|
-4.717
|
-7.551
|
-3.679
|
|
(4,094)
|
(4,263)
|
(5,072)
|
(0.0302)
|
(0.0311)
|
(0.0326)
|
(7.011)
|
(6.583)
|
(9.013)
|
|
sng_popn
|
-291.8**
|
-855.2***
|
-504.4***
|
-0.00426***
|
-0.0116***
|
-0.00579***
|
0.486*
|
1.573***
|
1.086***
|
|
(133.3)
|
(289.0)
|
(152.6)
|
(0.00121)
|
(0.00222)
|
(0.00129)
|
(0.292)
|
(0.467)
|
(0.270)
|
|
sng_area
|
97.04
|
91.45
|
-65.16
|
0.000610
|
0.00106
|
-0.000264
|
-0.518***
|
-0.382*
|
-0.497**
|
|
(146.6)
|
(137.8)
|
(125.9)
|
(0.00111)
|
(0.00107)
|
(0.00104)
|
(0.187)
|
(0.227)
|
(0.214)
|
|
fiscal_indep*governance
|
|
812.5***
|
|
|
0.00446***
|
|
|
-0.540**
|
|
|
(165.1)
|
|
|
(0.00124)
|
|
|
(0.227)
|
|
|
sng_popn*governance
|
|
17.86***
|
|
|
0.000203***
|
|
|
-0.0266**
|
|
|
(6.722)
|
|
|
(5.29e-05)
|
|
|
(0.0115)
|
|
|
sng_area*governance
|
|
-4.601
|
|
|
-5.12e-05
|
|
|
-0.00107
|
|
|
(4.357)
|
|
|
(3.69e-05)
|
|
|
(0.00600)
|
|
|
fiscal_indep*lowinc
|
|
|
-26,005***
|
|
|
-0.0596
|
|
|
18.37
|
|
|
(7,574)
|
|
|
(0.0540)
|
|
|
(13.30)
|
|
sng_popn*lowinc
|
|
|
-813.6***
|
|
|
-0.00624***
|
|
|
2.008***
|
|
|
(258.6)
|
|
|
(0.00227)
|
|
|
(0.467)
|
|
sng_area*lowinc
|
|
|
1.739
|
|
|
0.000923
|
|
|
-0.724*
|
|
|
(227.9)
|
|
|
(0.00191)
|
|
|
(0.408)
|
|
governance
|
84.94***
|
119.7***
|
78.36***
|
0.000535***
|
0.000802***
|
0.000481***
|
-0.0935***
|
-0.129***
|
-0.103***
|
|
(22.12)
|
(21.10)
|
(21.09)
|
(0.000170)
|
(0.000165)
|
(0.000157)
|
(0.0282)
|
(0.0290)
|
(0.0278)
|
|
lowinc
|
|
|
-6,140***
|
|
|
-0.0517***
|
|
|
7.770***
|
|
|
|
(1,001)
|
|
|
(0.00738)
|
|
|
(1.424)
|
|
|
|
|
|
|
|
|
|
|
|
Observations
|
237
|
237
|
237
|
237
|
237
|
237
|
237
|
237
|
237
|
|
R-squared
|
0.749
|
0.777
|
0.798
|
0.814
|
0.826
|
0.851
|
0.666
|
0.680
|
0.736
|
|
Notes
|
1. Heteroskedasticity-robust SEs in parentheses
2. *** p<0.01, ** p<0.05, * p<0.1
3. Results are robust to the use of
cluster-robust SEs
4. Control variables included: year fixed
effects, island group fixed effects, governance, educ, and urban
5. Results are robust to adding bankdep
and provrevpc among the controls
|
1. Heteroskedasticity-robust SEs in parentheses
2. *** p<0.01, ** p<0.05, * p<0.1
3. Results are robust to the use of
cluster-robust SEs
4. Control variables included: year fixed
effects, island group fixed effects, governance, educ, and urban
5. Results are robust to adding bankdep
and provrevpc among the controls
|
1. Heteroskedasticity-robust SEs in parentheses
2. *** p<0.01, ** p<0.05, * p<0.1
3. When cluster-robust SEs are used: a) sng_area
and most of the significant interaction terms became insignificant; b) fiscal_indep
remains insignificant
4. Control variables included: year fixed
effects, island group fixed effects, governance, educ, and urban
5. When bankdep and provrevpc were
included among the controls: a) about half of the significant
decentralization variables and their interaction terms became insignificant;
b) fiscal_indep remains insignificant
|
|
Variable
|
Description
|
|
Mean
|
Std. Dev.
|
Observations
|
|
per capita income
|
Per capita income (PPP) in
2012 Metro Manila Philippine Peso (PhP)
|
overall
|
40,772
|
12,173
|
N = 240
|
|
between
|
|
11,652
|
n = 80
|
|
within
|
|
3,678
|
T = 3
|
|
hdi
|
Human development index
|
overall
|
0.53
|
0.11
|
N = 240
|
|
between
|
|
0.11
|
n = 80
|
|
within
|
|
0.03
|
T = 3
|
|
poverty
|
Population poverty
incidence (in %)
|
overall
|
36.00
|
14.99
|
N = 240
|
|
between
|
|
14.28
|
n = 80
|
|
within
|
|
4.76
|
T = 3
|
|
fiscal_indep
|
Ratio of provincial
government own-sourced revenue to provincial government expenditures
|
overall
|
0.18
|
0.16
|
N = 240
|
|
between
|
|
0.15
|
n = 80
|
|
within
|
|
0.06
|
T = 3
|
|
sng_popn
|
Number of local
government units (cities and municipalities) per 100 thousand population
|
overall
|
3.38
|
4.18
|
N = 240
|
|
between
|
|
4.19
|
n = 80
|
|
within
|
|
0.23
|
T = 3
|
|
sng_area
|
Number of local
government units (cities and municipalities) per 1,000 square kilometers
|
overall
|
6.83
|
5.20
|
N = 240
|
|
between
|
|
5.22
|
n = 80
|
|
within
|
|
0.00
|
T = 3
|
|
governance
|
Good governance index
|
overall
|
123.81
|
23.36
|
N = 237
|
|
between
|
|
23.46
|
n = 79
|
|
within
|
|
0.00
|
T = 3
|
|
educ
|
Mean years of schooling
|
overall
|
8.24
|
1.11
|
N = 240
|
|
between
|
|
1.03
|
n = 80
|
|
within
|
|
0.44
|
T = 3
|
|
urban
|
Percent of population
living in urban areas
|
overall
|
25.49
|
22.96
|
N = 240
|
|
between
|
|
22.97
|
n = 80
|
|
within
|
|
1.95
|
T = 3
|
|
bankdep
|
Total bank deposits in
millions PhP
|
overall
|
15,861.98
|
29,186.78
|
N = 239
|
|
between
|
|
28,184.31
|
n = 80
|
|
within
|
|
7,847.41
|
T = 2.99
|
|
provrevpc
|
Provincial government
total revenue per capita
|
overall
|
1,293.80
|
1,357.46
|
N = 240
|
|
between
|
|
1,305.74
|
n = 80
|
|
within
|
|
389.86
|
T = 3
|
|
|
fiscal_indep
|
sng_popn
|
sng_area
|
governance
|
educ
|
urban
|
bankdep
|
provrevpc
|
|
fiscal_indep
|
1.000
|
|
|
|
|
|
|
|
|
sng_popn
|
-0.266
|
1.000
|
|
|
|
|
|
|
|
sng_area
|
0.183
|
0.521
|
1.000
|
|
|
|
|
|
|
governance
|
0.092
|
0.344
|
0.406
|
1.000
|
|
|
|
|
|
educ
|
0.563
|
0.165
|
0.509
|
0.298
|
1.000
|
|
|
|
|
urban
|
0.518
|
-0.371
|
-0.049
|
0.000
|
0.283
|
1.000
|
|
|
|
bankdep
|
0.476
|
0.000
|
0.288
|
0.202
|
0.627
|
0.458
|
1.000
|
|
|
provrevpc
|
-0.160
|
0.911
|
0.462
|
0.423
|
0.236
|
-0.299
|
0.028
|
1.000
|
Table 1Summary of utility functions and constraints DISPLAY Table
Table 2Regression results DISPLAY Table
Table A1Summary statistics DISPLAY Table
Table A2Correlation table DISPLAY Table
* The author thanks Alvin Ang, Fernando Aldaba, Rosalina Tan, Noel De Guzman, Agham Cuevas, Philip Arnold Tuaño, and Victor Venida for comments on an earlier version; and two anonymous referees whose comments helped to substantially improve this paper. All remaining errors are the author’s sole responsibility.
1 Like in the model without governance, the optimal values kD*, fD*, kC* and fC* can also be computed, although they are not needed in determining UD* and UC*. These are:  ; 
2 In the case where /L/ > /K/, the terms inside the bracket in equation (35) will be negative. Since it will be raised to a fraction (1- β), J could either be positive, negative, or imaginary.
3 Like in the model without governance, the welfare loss from adopting the wrong de/centralization policy can be computed. If the welfare maximizing policy is to decentralize, welfare loss from centralization is   . If centralization is welfare maximizing, the welfare loss from decentralization is  .
4 A second indicator – own-sourced revenues of the provincial government expressed as share of total revenue of the provincial government – was considered. However, it was no longer used because it is highly correlated with fiscal_indep (correlation coefficient of 0.92).
5 It must be noted that the variable fiscal_indep was computed using own revenues and expenditures of the provincial government – and not the consolidated budget of provincial plus sub-provincial governments.
6 I thank one anonymous reviewer for suggesting the inclusion of this independent variable.
7 Dummy =1 if provincial per capita income is below the median; =0 otherwise.
8 The 80 Philippine provinces are grouped into 16 so-called regions. These regions are not local governments but are merely geographic groupings.
|
|
March, 2021
I/2021
|
