Does political competition matter for public goods provision? Evidence from Russian regions Olga Vasilyeva [email protected] Amur State University
Mar 31, 2015
Does political competition matter for public goods provision?
Evidence from Russian regions
Olga [email protected] State University
• Does political competition matter for public policies under conditions of unfair elections and autocracy?
• Which mechanisms of government accountability give better results under conditions of suppressed political competition? – Formal vs informal
What does the literature tell?
• Topic 1: Political competition – Besley and Burgess (2002) – Careaga and Weingast (2001) – Acemoglu and Robinson (2006) – But implicit assumption about democracy!
• Topic 2: Efficiency of accountability mechanisms – formal vs informal– Sonin (2010) – Zhuravskaya and Persson (2011)
Stylized factsStylized facts: Russian regions
• 90s- early 2000s Political monopoly of local elite: weak central government and political parties
• mid-2000s Transition from the monopoly of local elite to the monopoly of federal elite
• Late 2000s Political monopoly of the federal elite: strong central government and the ruling party- “United Russia”
Stylized factsStylized facts: Russian regions
• 2001 - the law regulating political parties’ activities – 44 political parties (2003)
– 7 political parties (2009)
• 2003 - the new elections law– majoritarian rule was replaced by a mixed-proportional rule or
proportional rule (2007)
• 2005 replacement of governors’ elections by their appointments
Stylized factsStylized facts: Russian regions
• Two kinds of political competition– Between local and federal elites
– Between political parties – still weak at regional level !!! • No platforms! Cares of private benefits only
• Two kinds of governors– New bureaucrats
• represent interests of national ruling party and federal elite
– Old bureaucrats – have two contracts:• Explicit with federal elite and national ruling party
• Implicit with local elite
Conceptual Framework
• A new governor
1
)(,Pr 0
eg
etPSsgSU NG
• An old governor
1
,,Pr 0
leg
gfmetPSslgSU OG
Conceptual Framework
G
Pu
blic
go
od
s
Share of the ruling party
Governor-new bureaucrat
Governor-old bureaucrat
S0 S1 S2 S0
Conceptual Framework
Tested hypotheses
OGNG
OGNG
OGNG
ggs
ggs
ggs
10
1
0
0
0
0
HypothesesHypotheses
• governors-“new bureaucrats” provide less public goods than “old bureaucrats” if the share of the ruling party is very small or very big
• governors- “old bureaucrats” provide less public goods than “new bureaucrats” if the share of the ruling party is not too small or big
Data and estimation strategyData and estimation strategyUnbalanced panel for 74 regions 2004-2009: 216
observations
– Fixed effects for regions and years• Linear specification
• Quadratic specification
– SUR
DataData
Dependent variable: Public goods provisionPublic goods provision
INPUTS- Education
- Public spending per capita in region
- Share of education in total public expenditures in region
- Health care- Public spending per capita in region
- Share of health care in total public expenditures in region
Source of Data: Rosstat
DataDataDependent variable: Public goods provisionPublic goods provision
OUTCOMES- Infrastructure
- share of paved roads - share of the paved roads with improved surface - density of paved roads
- Health care- infant mortality rate - total mortality rate
- Education- Score for Final state exam – not yet
Source of Data: Rosstat
DataDataDependent variable: Public goods provisionPublic goods provision
OUTPUTS- Education
- number of students per teacher - number of students per class - number of computers per 100 students
- Health care- number of inhabitants per doctor - number of inhabitants per a bed in public hospitals
Source of Data: Rosstat
BUT…BUT…
• Changing the share of spending on a public good might be due to increase of other kinds of spending
• Observations only for few years – It is hard to believe that mortality rate or quality of
education might be affected by policies for 2- 3 years
• Outputs as well as outcomes might be determined not by regional policies but federal policies – in Russia public health care is financed by regional
governments as well as federal government
DataDataIndependent variables: Political variablesPolitical variables
• Dummy for “new” bureaucrat/”old” bureaucrat
• Share of the ruling party among representatives elected by party list in regional legislature (S)
– Logarithm of the Herfindahl–Hirschman Index (HHI)
• Linear interaction term between the dummy and the share of the ruling party
– Linear interaction term between the dummy and HHI
• Quadratic interaction term between the dummy and the share of the ruling party
– Linear interaction term between the dummy and HHI
Source of Data: Central Election Commission of the Russian Federation, The Inter-Regional Electoral Network of Assistance in Russia
Political variables: Political variables: Elections in Elections in regional legislaturesregional legislatures
9
26
42
6374
0
20
40
60
80
2004 2005 2006 2007 2008
The share of the ruling partyThe share of the ruling party: Mean 50%
Max 92% (Tatarstan in 2004 -2008)
Min 17% (Amur region in 2005-2007)
Political variables: Political variables: GovernorsGovernors
74
5138
15 9
17
24
4038
1219
27
6
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
2004 2005 2006 2007 2008
elected governors old bureaucrats new bureaucrats
DataData
Independent variables: ControlsControls– Log total budget spending per capita
– Share federal budget transfers in regional budget
– Household income per capita
Source of Data: Rosstat
– Governors’ characteristics• Age
• Duration of keeping office
• Outsiders (non resident)/ insiders (resident)
• Politicians / businessmen
Source of Data: collected by author
Governors’ characteristicsGovernors’ characteristics
The share of the governors-outsiders
9% 10%16% 19% 22% 24%
30%34%
0%
10%
20%
30%
40%
2003 г. 2004 г. 2005 г. 2006 г. 2007 г. 2008 г. 2009 г. 2010 г.
The share of the governors with mostly political background (not businessmen)
92% 89% 81% 80% 78% 80% 78% 75%
0%
20%
40%
60%
80%
100%
2003 г. 2004 г. 2005 г. 2006 г. 2007 г. 2008 г. 2009 г. 2010 г.
Average duration of keeping office
66,5
77,5
8
2003 2004 2005 2006 2007 2008 2009 2010
years
Average age of governors
52,0
52,5
53,0
53,5
54,0
54,5
55,0
55,5
2003 2004 2005 2006 2007 2008 2009 2010
age
(yea
rs)
SpecificationSpecification
Yit - measure of public goods provision
Si(t-k) - share of the ruling party NGi(t-k) - dummy for governors-”new bureaucrats”
Xi(t-1) - set of control variables
αi and ρt - regions and year fixed effects
The effect of governors –”new bureaucrats” is defined by:
ittiktiktiitktiktiktitiit XSNGSNGSNGY 152
4321
2431 SS
Results: Quadratic specifications
spending per capita on share of spending on
health care
educationhealth care
education
Dummy for a new governor0.32
(1.59)
-6.50*
(3.61)
7.01**
(3.12)
3.39
(2.35)
Dummy for a new governor x share of the ruling party
1.57
(7.63)
43.03***
(17.28)
-24.27
(14.95)
-15.94
(11.26)
Dummy for a new governor x share of the ruling party2
-5.71
(7.42)
-52.63***
(16.82)
18.45
(14.55)
11.42
(10.96)
Time fixed effect yes yes yes yes
Regional fixed effect yes yes yes yes
Number of observations 216 216 216 216
Results: Linear specifications
spending per capita on share of spending on
health care education health care education
Dummy for a new governor
1.39
(0.76)
3.40*
(1.79)
3.54**
(1.50)
1.24
(1.13)
Dummy for a new governor x share of the ruling party
-4.22***
(1.28)
-10.28***
(3.00)
-5.58**
(2.52)
-4.37**
(1.89)
Time fixed effect yes yes yes yes
Regional fixed effect yes yes yes yes
Number of observations 216 216 216 216
Results: Quadratic specifications
spending per capita on share of spending on
health care
educationhealth care
education
Dummy for a new governor0.32
(1.59)
-6.50*
(3.61)
7.01**
(3.12)
3.39
(2.35)
Dummy for a new governor x share of the ruling party
1.57
(7.63)
43.03***
(17.28)
-24.27
(14.95)
-15.94
(11.26)
Dummy for a new governor x share of the ruling party2
-5.71
(7.42)
-52.63***
(16.82)
18.45
(14.55)
11.42
(10.96)
Time fixed effect yes yes yes yes
Regional fixed effect yes yes yes yes
Number of observations 216 216 216 216
Max 41%
20% - 63%
Results: Outputs and outcomes
total
mortality rate
share of the paved roads
with improved
surfaces
students per class
Dummy for a new governor1.67***
(0.56)
1.80
(1.30)
-2.06*
(1.11)
Dummy for a new governor x share of the ruling party
-9.67***
(2.68)
-6.21***
(2.19)
6.62***
(2.25)
Dummy for a new governor x share of the ruling party2
10.63***
(2.61)- -
Time fixed effect yes yes yes
Regional fixed effect yes yes yes
Number of observations 216 216 216
Results: Outputs and outcomes
total
mortality rate
share of the paved roads
with improved
surfaces
students per class
Dummy for a new governor1.67***
(0.56)
1.80
(1.30)
-2.06*
(1.11)
Dummy for a new governor x share of the ruling party
-9.67***
(2.68)
-6.21***
(2.19)
6.62***
(2.25)
Dummy for a new governor x share of the ruling party2
10.63***
(2.61)- -
Time fixed effect yes yes yes
Regional fixed effect yes yes yes
Number of observations 216 216 216
Max 45 %
23% - 68%
Further workFurther work
• Reverse causality – IV approach
• a share of the ruling party in federal elections 2003, 2007 in a region;
• Putin/Medvedev’ share in President’s election in 2004, 2008
• Spending on infrastructure