Document History

Original Publish Date: 11 July, 2020

Updated on: 27 April, 2021


In this post we try to find how and what the characteristics of constituencies are associated with the dynasticism. We use census data and s and Ashers development data to explore the same.

Asher

Asher data provides us information at the assembly constituency level for the the Indian states. The dependent variable corresponds to the number of years rules by a dynast ruler in a particular constituency in a 30 year period starting 1970 to 2001. We limit our analysis to this period because this data is from the populations census and economic census which were conducted in early 2000s. We use the 91 population census as a base given the lack of many variables in the 1971 popultaion census data provided by Francesca.

All the models is a poisson regression regressions. In Asher’s data set all the variables have a urban and rural counter parts. The first one, we have all urban and rural variables in one model. Second and third ones are separate urban and rural models respectively.

Regression Results - Dev Asher
Dependent variable:
Dynast rule
(1) (2) (3)
pc91_vd_power_supl -0.692*** -0.755***
(0.207) (0.205)
el_con91 0.00000* 0.00000
(0.00000) (0.00000)
p_sch_r91 0.100 -0.073
(0.293) (0.285)
p_sch_u91 -0.010** -0.010**
(0.005) (0.005)
s_sch_u91 -0.002 -0.003
(0.007) (0.006)
irr_share91 -0.262 -0.400 -0.150
(0.311) (0.301) (0.299)
hosp_r91 -12.787*** -13.343***
(2.235) (2.250)
hosp_u91 0.027 0.030*
(0.018) (0.018)
pc91u_td_p_road 0.0002 -0.0004
(0.002) (0.002)
pc91_vd_app_pr 0.260 0.291
(0.279) (0.278)
constituency_typeSC -1.096*** -1.041*** -1.099***
(0.130) (0.129) (0.129)
p_urban 0.010*** 0.009***
(0.003) (0.003)
Prural -0.010***
(0.003)
Constant 1.363*** 1.888*** 1.455***
(0.230) (0.234) (0.224)
Observations 359 359 359
Log Likelihood -1,154.276 -1,180.365 -1,161.160
Akaike Inf. Crit. 2,334.553 2,378.730 2,338.320
Note: p<0.1; p<0.05; p<0.01

Francesca

We use Francesca’s ac level census data to regress the change in the constituency characteristics against the years ruled by dynast within a 30 year period(1971-2001). In this poisson regression we used sub-regions as controls.

Regression Results - Dev Francesca
Dependent variable:
Dynast rule
lit -0.390***
(0.091)
work -0.841**
(0.346)
agr_work -0.323
(0.239)
constituency_typeSC -1.037***
(0.124)
Observations 394
Log Likelihood -1,275.333
Akaike Inf. Crit. 2,572.666
Note: p<0.1; p<0.05; p<0.01

ADR & Census

In this section we use the All India GE level dynast data to asses the characteristics of constituencies where dynasts win or contest. This is a probit regression with state and year fixed effects.

Dependent variable:
dynast_winner dynast_candidate
(1) (2)
turnout -0.003 0.004
(0.005) (0.004)
enop -0.324*** -0.174***
(0.052) (0.043)
reservationSC/ST -0.145* -0.175***
(0.077) (0.066)
literacy 0.070 -0.236
(0.502) (0.433)
urban_pop -0.428* -0.283
(0.228) (0.198)
muslim_pop 0.144 0.347
(0.264) (0.227)
sc_st_pop 0.301 0.425
(0.335) (0.289)
agri_lab -1.113 -1.615
(1.194) (1.057)
Observations 3,900 3,900
Log Likelihood -1,236.354 -1,761.517
Akaike Inf. Crit. 2,560.707 3,611.034
Note: p<0.1; p<0.05; p<0.01

HIL Wealth index

We regressed the wealth score which was calculated from the 20011 census against the number of years ruled by dynast in a constituency from 1977 to 2004.

Dependent variable:
dyn_rule
wealthscore 0.039***
(0.006)
constituency_typeSC -1.471***
(0.243)
Constant 0.472**
(0.213)
Observations 81
Log Likelihood -345.751
Akaike Inf. Crit. 697.502
Note: p<0.1; p<0.05; p<0.01