do electoral politics matter in mgnrega implementation ... · 13 to 17 days more nregs work compare...
TRANSCRIPT
![Page 1: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/1.jpg)
Do Electoral Politics Matter in MGNREGA Implementation?
Evidence from Village Council Elections in West Bengal
Subhasish Dey, University of Manchester
Kunal Sen, University of Manchester
4th West Bengal Growth Workshop,
Indian Statistical Institute, Kolkata
27th December 2014
![Page 2: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/2.jpg)
Outline
• Motivation
• Research Objectives & question(s)
• The specific context
• Data
• Empirical Methodology
• Results
• Conclusion
![Page 3: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/3.jpg)
Motivation
• Political economy of redistributive politics
(Bardhan and Mookherjee 2010a, 2010b; Benhabib and
Przeworski 2006; Dixit 1996; Dixit and Londregan 1996)
• Feedback effect of politics of redistribution on electoral outcome.
(Finan and Schechter 2012; Litschig and Morrison 2012; Werker
et al. 2012)
![Page 4: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/4.jpg)
Research Objectives & questions
Objective:
1) Whether political nepotism exists in provisioning of Public work
2) Whether political nepotism is electorally rewarding.
“Political Nepotism” as a situation when existing ruling party in a
democratic govt. positively discriminates its own party constituencies
from other party constituencies in allocating public fundfrom other party constituencies in allocating public fund
Main Research Question
1) Does Village Council (Gram Panchayat) ruling party discriminate
between constituencies in delivering public fund(NREGS)?
2) If yes (or not), what are the political feedback effect in the next
election outcome of the previous ruling party?
![Page 5: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/5.jpg)
The specific context
• Village Council Election (Gram Panchayat election) in India
Village Council
(Gram Panchayat-GP)
Village Council Chairman
Ruling party: Right Populist
4 out 6
4
2
Village Council spends annually
£250K-£300K on developmental
programme. NREGS constitutes
85-90% of that spending.
Ward-1
(Gram Sansad-GSWard-6
Ward-2 Ward-3 Ward-4 Ward-5
Right-Populist Right-Populist Left Right-Populist Right-PopulistLeft
Ward level Winning party
![Page 6: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/6.jpg)
• Self-Selected
Programme
• Budget $ 7 billion (0.6%
of GDP, India) per year
• Covering 50 million
households per year
What is MG-NREGS?
Covering 50 million
households per year
• Village council/GP is the
PIA
• Village Chairman is the
key person in
implementation
• Social Audit
![Page 7: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/7.jpg)
INDIAWhere I did my
survey
Year Purulia South 24 Parganas Jalpaiguri
2008 Left Right Populist Left
2013 Right Populist Right Populist Marginal Left
![Page 8: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/8.jpg)
Data• Detailed Village Council election results:2008 and 2013
• Total 569 wards (or village/gram sansad) over 49 Village councils from 24
Blocks under 3 districts in West Bengal, India.
• Ward level NREGS info (expenditure, no. of schemes, no. of household
participated) and other detail info on other developmental schemes for
2010, 2011, 2012
• ward level rain fall data
• Ward level socio-economic-demographic info.• Ward level socio-economic-demographic info.
Source:
• Election Results from West Bengal State Election Commission.
• Primary survey of 569 wards/villages & 49 GPs to get 2010-2012 panel data.
• Census 2011, Govt. of India and Rural Household Survey 2012, Govt. of WB.
• Latitude-Longitude wise monthly rain fall data from Centre for Climate
Research at the University of Delaware and National Climatic data centre.
3 years village level panel data, 2010-2012, correspondingly to election year
2008 and 2013
![Page 9: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/9.jpg)
Some Trend: Political Scenario in West Bengal
89.49
75.6189.67 87.33 88.27
86.82
69.25
25.7
10.35
22.88
9.27 10.99 9.78
11.64
29.28
74.29
10
20
30
40
50
60
70
80
90
100
Sha
re o
f sea
t (in %
)
10.35
0
10
1978 1983 1988 1993 1998 2003 2008 2013
Left front share of seat Congress share of seat
TMC share of seat Congress & TMC share of seat
Year
Seat share of major political parties in ZillaParishad (i.e. the district level tier of the local government) Election over the years
![Page 10: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/10.jpg)
Some Trend: Political Scenario in West Bengal
2003 2008 2013
District wise ruling party position after the Local Government Elections
CPIM: Congress: TMC:
![Page 11: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/11.jpg)
Descriptive results
2010-2012 NREGS Expenditure and ward level winning party after 2008 election
Party wise winning
seat allocation
![Page 12: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/12.jpg)
Expecting Jump in P(T=1) and the on the outcome
Perfect compliance: Sharp RDPerfect compliance: Sharp RD
Imperfect compliance: Fuzzy RD
![Page 13: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/13.jpg)
Graphical analysis: Jump in % of ruling-party winning
candidate
![Page 14: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/14.jpg)
Graphical analysis: Jump in value in outcome varaible
![Page 15: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/15.jpg)
![Page 16: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/16.jpg)
Local Average Treatment Effect (LATE): Wald Estimator
]|[lim]|[lim
]|[lim]|[lim
00
00
εεεεσ
εε
εε
+=−+=+=−+=
=↑↓
↑↓
cXTEcXTE
cXYEcXYEFRD
![Page 17: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/17.jpg)
Trying to find the causal effect of Treatment on
Outcome. Some terminology before FRDD
• Treatment: A dummy (T): either ‘0’ or ‘1’
• T=1: when a village council/GP ward (or simply village) is a
ruling party ward.
• T=0: Otherwise.
Outcome (Y): Ward/Village level NREGS outcome (namely NREGS
Expenditure and NREGS days availed by a household)
Assignment/forcing variable(X): Village wise GP level ruling
party’s vote share after 2008 Panchayat Election.
![Page 18: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/18.jpg)
Empirical Methodology
• We used Fuzzy Regression Discontinuity Design (FRDD).
• Our base line specification
eTXfY ++= σ)(
Where
= Local average treatment effect (LATE) on outcome variable Yσ = Local average treatment effect (LATE) on outcome variable Y
(shows the effect of being ruling-party winning-member on
sansad wise NREGS expenditure)
e = other unobserved error
We are concerned to find sign, magnitude and statistical significance of T.
But ‘T’ is endogenous. Unobserved local factors explaining T can explain Y
directly i.e. E(T,e)≠0 and hence is not idenRfied.σ
![Page 19: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/19.jpg)
Empirical Methodology
We would like to see whether there is any discontinuity in outcome variable
following the discontinuity in probability of Treatment.
If >0 => there will be a upward jump in the E(Y|X) at the X=50 meaning
Village Council Ruling party wards systematically have higher NREGS
expenditure compare to opponent party wards.
σ
![Page 20: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/20.jpg)
Control variables
Since NREGS is a demand driven programme, we control for
demand side factors and we also control ward level winning
member’s characteristics.
Controls on demand side factors at the ward/village:
average monsoon rain fall, total voters in ward, total number of
households, number of BPL households, worker-to non-worker households, number of BPL households, worker-to non-worker
ratio, no. of minority households.
Controls on ward level winning member character:
Sex, Caste,
District dummy, year dummy
![Page 21: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/21.jpg)
Estimation Results
![Page 22: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/22.jpg)
Estimation Results
![Page 23: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/23.jpg)
Estimation Results
![Page 24: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/24.jpg)
Estimation Results
![Page 25: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/25.jpg)
Test for Validity of FRD
• Sensitivity analysis with different bandwidth and different
order of Polynomial.
• Sensitivity of Treatment effect with the inclusion of all
covariates
• Checking discontinuity of covariates at cut-off point. • Checking discontinuity of covariates at cut-off point.
• Density plot of forcing Variable
• Placebo test or falsification test: Checking discontinuity in
non-discontinuity point.
![Page 26: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/26.jpg)
Discontinuity of covariates at cut-off point
![Page 27: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/27.jpg)
Density of plot of forcing Variable
![Page 28: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/28.jpg)
Placebo test or falsification test: Checking
discontinuity in non-discontinuity point
![Page 29: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/29.jpg)
Findings on Causal effect of Treatment on Outcome
• Ruling party spends around INR 40K-50K more NREGS funds in their own village compare to opponents village.
• Household in the ruling party’s village gets 4 to 4.5 days more NREGS work compare to non-ruling party village.
• When TMC is the ruling party they spends 125K to 150K more NREGS funds in their own village compare to opponents village
• When TMC is the ruling party, household in the ruling party village gets 13 to 17 days more NREGS work compare to household in a non-ruling 13 to 17 days more NREGS work compare to household in a non-ruling party’s village.
• When CPIM is the ruling party they spends around 20K less NREGS funds in their own party village but these results are statistically insignificant.
• When CPIM is the ruling party, household in the ruling party village gets 2 to 3 days less NREGS work compare to household in a non-ruling party’s village.
![Page 30: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/30.jpg)
Results on Reciprocity or feedback effect
![Page 31: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/31.jpg)
Results on Reciprocity or feedback effect
![Page 32: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/32.jpg)
Empirical strategy for the feedback effect on 2013 election
Outcome:
)8.......()()(0 exTExfY ++= σ
• This was our treatment effect Equation. We use predicted value of Y for T=1 from the above equation as our main explanatory variable to get causal effect of discrimination on following election outcome.
• Then we run following regression with OLS.
ii KhatYV εγαα +++= _102013_
• Where is the 2008 ruling party’s vote share in 2013 panchayat
election at village i,
• is the predicted value of Y from equation 8 above
• K is vector of other village level controls including margin of win in the
2008 election .
• Essentially this is also a IV estimation where treatment (T) is the
instrument
2013_iV
hatY _
![Page 33: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/33.jpg)
Results on Feedback Effect
![Page 34: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/34.jpg)
Results on Feedback Effect
![Page 35: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/35.jpg)
Conclusion
• We tried to look whether Political Nepotism operates at the local
govt. level and if so then what is its feedback effect on the
following election outcome.
• We find overall evidence of discrimination in village wise
expenditure on Public Good (NREGS).
• Village Council level Ruling party spends more in own party • Village Council level Ruling party spends more in own party
constituency. However, results differ between specific parties
• Right Populist Party reaping out significant benefit in 2013
election through this behaviour of Political Nepotism .
• Left does not came to behave in this way for which they pay in
loosing vote share and lower probability of re-elected 2013
![Page 36: Do Electoral Politics Matter in MGNREGA Implementation ... · 13 to 17 days more NREGS work compare to household in a non -ruling party’s village. • When CPIM is the ruling party](https://reader033.vdocuments.us/reader033/viewer/2022060508/5f232121c25d8836fd42f68e/html5/thumbnails/36.jpg)
Contribution to the literature:
“……..this study tries contribute an alternative explanation of re-
election motive of the incumbent ruling party and this
explanation is certainly a deviation from the existing theory that
given the predictions of standard voting models, which says
political leaders who are concerned with re-election would focus
on delivering benefits to ‘swing voters’ and not the loyalists..”
Thank You
on delivering benefits to ‘swing voters’ and not the loyalists..”
But why do ‘Left’ and ‘TMC’ behave differently in allocating
NREGS?? Future work