THE POLITICS OF FOOD DISTRIBUTION IN INDIA A R N A V M O D Y I N - - PowerPoint PPT Presentation

A R N A V M O D Y I N T E R N A T I O N A L R E L A T I O N S H O N O R S T H E S I S

THE POLITICS OF FOOD DISTRIBUTION IN INDIA

Topic

• Distributive Politics and the Political Business Cycle
• The influence of politics and elections on food

distribution to Indian states through the Public Distribution System (PDS)  States that are aligned and swing benefit  States that are aligned and have a greater representation in parliament benefit

• Costs $13.6 billion a year

(~1% of GDP)

• 29 states
• Elections every five years
• Single member constituency

first past the post majoritarian system

Farmer

• Procurement of grains

from farmers

Center

• Allocation to states

based on number of families

State

• Distribution to fair price

shops within states

Shops

• Sale of subsidized

grains to eligible beneficiaries

Introduction

Literature Review and Theory

Voters represented by members of a majority party receive more benefits Gordon and Kang 2015 Political Business Cycles in India: Fiscal policies and pre- election increase in budget deficit Khemani 2004 Ration shops are crucial in meeting daily food needs for two-thirds of the population The central incumbent has discretionary influence in PDS A perception that pre-election policies will continue post- election

Literature Review and Theory

Goodwill leakages and alignment Arulampalam et al. 2009 Aligned Swing Effect Arulampalam et al. 2009 Leakage of goodwill (credit) to state government from central government policies States that are swing and/or have a larger representation in parliament are deemed more important

Hypotheses

Hypothesis 1: Aligned Swing Effect

• States that are aligned with the central government and are swing

receive a higher percentage offtake and more ad-hoc allocations

• f wheat and rice

Hypothesis 2: Political Business Cycle

• States experience a higher percentage offtake closer to elections
• States that are aligned and swing receive a higher percentage
• fftake closer to elections

Hypothesis 3: Importance

• States with a larger number of national constituencies receive a

higher percentage offtake and more ad-hoc allocations of wheat and rice

Summary Statistics

Name (units) Mean Std Dev Min Max Source # Obs

• fftakest (%)

91.43% 25.91 0% 298.60% IndiaStat 1,667 adhocst 0.32 0.47 1 IndiaStat 1,667

alliedst

0.47 0.50 1 ECI 1,667

monthsst

30.94 18.5 74 ECI 1,667

netswingst

0.38 0.49 1

• 1,632

stateswingst

0.47 0.50 1

• 1,632

centralswingst

0.28 0.45 1

• 1,632

close_3st

0.07 0.25 1

• 1,667

close_6st

0.12 0.319 1

• 1,667

close_12st

0.22 0.41 1

• 1,667

seatsst

18.9 18.88 1 80 ECI 1652

incomest (INR)

75,290 39,698 11,051 224,138 RBI 1,652

Hypothesis 1: Aligned Swing Effect

OLS (1)

• fftakest

OLS (2)

• fftakest

OLS (3)

• fftakest

OLS (4) adhocst Probit (5) adhocst Probit (6) adhocst Probit (7) adhocst alliedst

• 1.18
• 4.00
• 2.02
• 0.03
• 0.08
• 0.13
• 0.10

(2.46) (2.41) (2.48) (0.05) (0.14) (0.17) (0.13)

netswingst

• 2.49
• 0.05
• 0.18

(3.83) (0.05) (0.16)

alliedst x netswingst 6.92* 0.13** 0.39**

(3.38) (0.05) (0.17)

stateswingst

• 12.22**
• 0.05

(0.040) (0.21)

alliedst x stateswing 16.17*** 0.31

(5.74) (0.27)

centralswingst

• 11.48*
• 0.55**

(5.62) (0.27)

alliedst x centralswing 8.59* 0.39*

(4.44) (0.21)

excess_rainfallst 0.49** 0.09 0.05 0.11

(0.18) (0.14) (0.15) (0.13)

log(incomest) 15.34 14.33 17.02 0.03 1.5** 1.41 1.58**

(11.09) (12.38) (10.33) (0.04) (0.65) (0.82) (0.64)

# Obs 1,652 1,620 1,601 1,652 1,593 1,561 1,542 R2 0.17 0.18 0.18 0.16 0.12 0.13 0.12 All models include state fixed effects, year fixed effects and clustered standard errors at the state level Robust standard errors in parentheses *p<0.10, **p<0.05, ***p <0.01

Coefficient Plots

The aligned swing effect is significant at the 90 percent confidence interval

Allied Non-Swing Non-Allied Swing Allied Swing

• 20
• 10

10 20 Allied Non-Swing Non-Allied Swing Allied Swing

• 10
• 5

5 10 15 Allied Non-Swing Non-Allied Swing Allied Swing

• 20
• 10

10 20 30

State Swing Central Swing Net Swing

Hypothesis 2: Political Business Cycle

OLS (8)

• fftakest

OLS (9)

• fftakest

OLS (10)

• fftakest

OLS (11)

• fftakest

OLS (12)

• fftakest

OLS (13)

• fftakest

OLS (14)

• fftakest

monthsst 0.06* 0.01

• 0.05

(0.073) (0.04) (0.30) months2

st

0.002 1.13 1.00 1.38 1.19 (0.006) (2.74) (2.55) (2.44) (2.75) alliedst

• 2.20
• 0.71

(3.14) (3.91) monthsst x alliedst 0.10

• 0.03

(0.08) (0.39) months2

st x alliedst

0.002 (0.006) close_3st

• 2.39
• 2.39
• 3.09

(4.38) (4.38) (3.21) close_6st

• 3.23
• 0.79
• 4.60

(3.98) (2.93) (2.41) close_12st 2.82 2.84 (2.84) (2.84) close_3st x alliedst

• 11.71*
• 11.69*
• 5.90

(5.83) (5.85) (5.72) close_6st x alliedst 6.09 6.18

• 2.31

(6.07) (4.78) (5.08) close_12st x alliedst 0.02

• 0.04

(4.12) (4.13) log(incomest) 1.64 1.94 1.79 2.13 1.99 2.01 2.01 (4.01) (4.23) (4.46) (4.07) (4.11) (4.11) (4.12) # Obs 1,667 1,667 1,667 1,652 1,652 1,652 1,652 R2 0.16 0.17 0.17 0.17 0.17 0.17 0.16

All models include state fixed effects and clustered standard errors at the state level Robust standard errors in parentheses *p<0.10, **p<0.05, ***p <0.01

Hypothesis 3: Importance

OLS (15)

• fftakest

OLS (16)

• fftakest

OLS (17)

• fftakest

OLS (18) adhocst Probit (19) adhocst alliedst

• 1.41
• 5.56**
• 3.65
• 1.15**
• 0.44**

(3.54) (2.07) (4.72) (0.07) (0.19)

seatsst

• 0.61***

(0.16)

alliedst x seatsst

0.18

(0.17)

log(seatsst)

• 4.83***
• 10.07***
• 0.19***
• 0.63***

(0. 57) (2.16) (0.03) (0.10)

alliedst x log(seatsst)

2.74*** 2.37 0.078*** 0.24***

(0.486) (1.94) (0.02) (0.07)

excess_rainfallst

13.83

• 2.80

13.21 0.43** 1.32**

(10.80) (1.64) (10.95) (0.18) (0.64)

log(incomest)

0.03 0.09

(0.04) (0.13)

year FE

    

state FE

   

cluster(state)

   

cluster(year)



# Obs

1,652 1,652 1,652 1,652 1,593

R2

0.17 0.04 0.17 0.16 0.12 Robust standard errors in parentheses *p<0.10, **p<0.05, ***p <0.01

Margins Plot (Model 18)

• .4
• .2

.2 .4 .5 1 1.5 2 2.5 3 3.5 4 4.5 logseats

Average Marginal Effects of Allied States With 95% CIs

OLS regression with adhocst as the dependent variable, state fixed effects, year fixed effects and standard errors clustered at the state level

Uttar Pradesh, Maharashtra, West Bengal, Bihar, Tamil Nadu, Madhya Pradesh, Karnataka, Gujarat, Andhra Pradesh, Rajasthan, Odisha, Kerala

Conclusion

States that are allied and swing receive higher offtakes and are more likely to receive ad-hoc allocations

• States that are not allied and swing receive lower offtakes

No substantial evidence of a political business cycle States that have more seats and are allied receive higher offtakes and are more likely to receive ad-hoc allocations

• Larger states that are not allied receive lower allocations of food

Two main theoretical notions:

• Some states are more important than others at different times
• Leaders can identify goodwill generation by policies and parties

to which voters will attribute this generated goodwill 1 2 3

Conclusion

States that are allied and swing receive higher offtakes and are more likely to receive ad-hoc allocations

• States that are not allied

and swing receive lower

• fftakes

No substantial evidence of a political business cycle States that have more seats and are allied receive higher offtakes and are more likely to receive ad-hoc allocations

• Larger states that are not

allied receive a lower allocation of food 1 2 3

APPENDIX

Additional Models: Aligned Swing Effect

OLS (20)

• fftakest

OLS (21)

• fftakest

OLS (22)

• fftakest

OLS (23) adhocst OLS (24) adhocst OLS (25) Adhocst

alliedst 0.16 0.72

• 0.81

0.21

• 0.05
• 0.85

(2.74) (2.77) (4.25) 4.02 (2.62) (2.66) netswing_05st

• 1.91

(3.62) alliedst x netswing_05 st 4.01 (4.15) netswing_10st

• 3.63

(3.57) alliedst x netswing_10st 2.40 (3.99) stateswing_02st 6.13 (5.38) alliedst x stateswing_02st 15.69 (13.76) stateswing_10st

• 12.24**

(5.78) alliedst x stateswing_10st 6.70 (5.75) centralswing_02st

• 4.94

(4.01) alliedst x centralswing_02st 3.33 (4.01) centralswing_05st

• 7.90

(5.72) alliedst x centralswing_05st 4.83 (5.28) log(incomest) 13.92 14.87 15.69 13.04 15.93 17.43 (10.69) (10.59) (13.76) (12.27) (10.88) (10.39) # Obs 1,649 1,652 1,620 1,620 1,652 1,601

All models include state fixed effects, year fixed effects and clustered standard errors at the state level Robust standard errors in parentheses *p<0.10, **p<0.05, ***p <0.01

Allied Non-Swing Non-Allied Swing Allied Swing

• 20
• 10

10 20 30

Coefficient Plots

The aligned swing effect is significant at the 95 percent confidence interval for state swing and net swing

Allied Non-Swing Non-Allied Swing Allied Swing

• 20
• 10

10 20 Allied Non-Swing Non-Allied Swing Allied Swing

• 10
• 5

5 10 15

State Swing Central Swing Net Swing

Margins Plot (Model 19)

Probit regression with adhocst as the dependent variable, state fixed effects, year fixed effects and standard errors clustered at the state level

• .2
• .1

.1 .2 .5 1 1.5 2 2.5 3 3.5 4 4.5 logseats

Average Marginal Effects of Allied States With 95% CIs

Further Considerations

• Increased government transparency would have

helped target the research

• Unmeasurable factors such as corruption and

inefficiency were not considered

• Electoral data collected from Indian election

between 2006 and 2015 can be applied to many

• ther areas of public policy and government

programs