0.0.1 Document History

Original Publish Date: 7th March, 2021

Updated on: 19 August, 2021

1 UP

The All India Survey on Higher Education (AISHE) website provided two kind of college data. 1, Colleges affiliated to Universities. 2, Stand alone institutions.

There were 7690 affiliated colleges 1136 stand-alone colleges. Among these two categories, only 7718 observations had year of establishment. Once I filtered the list of colleges that provided the year of establishments to colleges that are established after 1974, the observations further came down to 7002. This again reduced to 6825 after geocoding and chucking out the universities.

1.1 Data summary

1.2 Main variables

##  [1] "District"        "University.Type" "University.Name" "inst_type"      
##  [5] "college_name"    "college_type"    "Address"         "Website"        
##  [9] "Management"      "year_estd"       "Specialised.in"  "Location"       
## [13] "Upload.Year"     "full_name"       "full_adress"     "lat"            
## [17] "long"            "AC_NO"           "AC_NAME"         "PC_NO"          
## [21] "PC_NAME"

1.3 types of management

## # A tibble: 5 x 2
##   Management         count
##   <chr>              <int>
## 1 Central Government    11
## 2 Local Body           344
## 3 Private Aided        314
## 4 Private Un-Aided    5801
## 5 State Government     429

1.4 Year by year break-up of number of colleges

I kept Private un-aided as private and clubbed all others as government colleges.

After my initial inspections, I went to merge the college data with UP dynasties data after allotting the coleges to each election year based on the year of establishment

## # A tibble: 12 x 2
##    year_el count
##    <chr>   <int>
##  1 1974       55
##  2 1980       64
##  3 1985       43
##  4 1987       40
##  5 1989       40
##  6 1991       50
##  7 1993      167
##  8 1996      554
##  9 2002     1239
## 10 2007     2050
## 11 2012     2297
## 12 2017      300

Average number of colleges built every year in a constituency is 1.37

1.5 Distribution of number of colleges in a constituency

This density chart shows the distribution of the number of colleges built in different constituencies during last 10 years

1.5.1 Post 2009

1.5.2 Pre 2009

1.6 regressions

1.6.1 Dynast definition 2

regression run on dynast definition 2

1.6.1.1 All years

In this regression we look at colleges built during the rule of MLAs selected during the election cycles starting from 1974 to 2017

fit.poisson fit.probit
Dependent Var.: n_colleges n_college_bin
dyn_cum_2 -0.0140 (0.1075) -0.2462* (0.1171)
term_duration -0.0682 (0.1139) -0.1124 (0.1596)
turnout_percentage -0.0081 (0.0082) -0.0072 (0.0071)
margin_percentage 0.0020 (0.0051) -0.0057 (0.0037)
constituency_typeSC -0.0863 (0.1964) 0.1361 (0.2613)
enop 0.0030 (0.0549) -0.0984* (0.0446)
log(electors) 0.1101 (0.3834) 0.1188 (0.4021)
Caste control Yes Yes
Fixed-Effects: —————- —————–
year_el Yes Yes
constituency_name Yes Yes
___________________ ________________ _________________
Family quasipoisson(“log”) Probit
S.E.: Clustered by: constituen.. by: constituenc..
Observations 4,275 4,177
Squared Cor. 0.65490 0.53951
Pseudo R2 – 0.47348
BIC – 7,228.6

1.6.1.2 Post 1991

  • Tried post 1991 and 1996. Dynast co-effient on both are negative. The model below is post 1991.
fit.poisson fit.probit
Dependent Var.: n_colleges n_college_bin
dyn_cum_2 -0.0765 (0.0961) -0.1872* (0.0824)
term_duration 0.0173 (0.0491) 0.0028 (0.0597)
turnout_percentage -0.0209*** (0.0054) -0.0050 (0.0046)
margin_percentage 0.0044 (0.0038) 0.0036 (0.0034)
enop -0.0310 (0.0435) -0.0524 (0.0434)
log(electors) 0.9001* (0.3659) 0.7062. (0.3942)
constituency_typeSC -0.1897 (0.1556) 0.0129 (0.1232)
Caste control Yes Yes
Fixed-Effects: ——————- —————–
year_el Yes Yes
district_name Yes Yes
___________________ ___________________ _________________
Family quasipoisson(“log”) Probit
S.E.: Clustered by: district_name by: district_name
Observations 2,811 2,811
Squared Cor. 0.36457 0.37467
Pseudo R2 – 0.31312
BIC – 3,407.7

1.6.1.3 1993-2012 with night lights

UP colleges - Poisson model
fit.poisson fit.probit
Dependent Var.: n_colleges n_college_bin
dyn_cum_2 -0.0258 (0.0975) -0.2317* (0.1034)
incumbent 0.0331 (0.0802) 0.0695 (0.1020)
margin_percentage 0.0036 (0.0046) 0.0014 (0.0054)
turnout_percentage -0.0169** (0.0063) -0.0072 (0.0067)
enop -0.0082 (0.0459) -0.0928. (0.0545)
term_duration 0.0536 (0.0831) -0.0440 (0.1232)
n_schools 0.0015*** (0.0002) 0.0006** (0.0002)
nl_tot 4.75e-5*** (1.24e-5) 2.49e-5 (1.52e-5)
constituency_typeSC -0.2069* (0.0886) 0.1032 (0.1539)
log(electors) 0.1581 (0.4339) 0.5690 (0.5972)
no_terms 0.0074 (0.0436) -0.0178 (0.0391)
caste_uc 0.1428 (0.1617)
caste_yadav 0.2264 (0.2420)
caste_non_yadav_obc 0.2650 (0.1841)
caste_dalit 0.1384 (0.1804)
caste_muslim 0.4485* (0.1984)
Fixed-Effects: ——————– —————–
year_el Yes Yes
district_name Yes Yes
___________________ ____________________ _________________
Family quasipoisson(“log”) Probit
S.E.: Clustered by: district_name by: district_name
Observations 1,717 1,717
Squared Cor. 0.38965 0.31895
Pseudo R2 – 0.27639
BIC – 2,294.9

1.6.1.4 adr variables

Inorder to use the adr variables in the model, I have to limit the years to post 2009.

fit.poisson fit.probit fit.probit_no_se
Poisson Probit Probit without SE Clustered
Dependent Var.: n_colleges n_college_bin n_college_bin
dyn_cum_2 -0.0900 (0.0393) -0.2089*** (0.0496) -0.1381*** (0.0245)
incumbent 0.0925** (0.0009) -0.0257 (0.2049) 0.0473 (0.1384)
margin_percentage 0.0012 (0.0064) 0.0221*** (0.0038) 0.0143*** (0.0025)
turnout_percentage -0.0255 (0.0132) -0.0006 (0.0064) 0.0197*** (0.0038)
enop 0.0174 (0.0172) 0.0524 (0.1039) -0.0589 (0.0694)
constituency_typeSC -0.0394 (0.1717) 0.1306 (0.3922) 0.0924 (0.0885)
log(electors) 1.361 (0.2712) 1.839*** (0.0125) 1.329*** (0.1409)
log(total_immovable_assets_totals) 0.0015 (0.0023) -0.0018 (0.0063) 0.0075 (0.0111)
log(total_movable_assets_totals) 0.0551 (0.0272) 0.0100 (0.0332) 0.0005 (0.0255)
serious_crime -0.0025 (0.0038) 0.0224 (0.0328)
non_serious_crime 0.0173. (0.0014) 0.0139 (0.0199) 0.0048 (0.0042)
no_terms 0.1036 (0.1031) 0.0586 (0.0800)
Fixed-Effects: —————– ——————- ——————-
year_el Yes Yes Yes
district_name Yes Yes No
__________________________________ _________________ ___________________ ___________________
Family quasipoisson(“log”) Probit Probit
S.E.: Clustered by: year_el by: year_el by: year_el
Observations 803 779 803
Squared Cor. 0.46917 0.38684 0.26312
Pseudo R2 – 0.33302 0.21573
BIC – 1,282.4 947.26

2 India

There were 45022 affiliated colleges 11987 stand-alone colleges. Among these two categories, only 52232 observations had year of establishment. This further reduced to 49959 observations after geocoding .

2.1 Data summary

2.1.1 Year by year break-up of number of colleges

2.1.2 state break-up

We are focusing on data post 2009 since we only have pan-India dynasty data after 2009. Thus in our college data set before mergin with dynasty has 19148 observations.

2.1.3 State wise break-up post 2009

Average number of colleges built in a constituency in India during last 10 years is 1.67

2.1.4 Distribution of number of colleges established every election year in a constituency

2.1.5 Distributions of the number colleges established in the constituencies during 2009&2014

2.2 regressions

All pan-india regressions are run on colleges built during 2009-19 at PC level.

2.2.1 All colleges

Regression Results - India all colleges
Dependent variable:
n_colleges n_college_binary
OLS normal glm: quasipoisson probit
link = log
(1) (2) (3) (4)
dyn -0.277** -0.277** -0.154*** -0.070
(0.117) (0.117) (0.045) (0.072)
Year fixed effects Yes Yes Yes Yes
State fixed effects Yes Yes Yes Yes
Observations 4,631 4,631 4,631 4,631
R2 0.288
Adjusted R2 0.282
Akaike Inf. Crit. 23,555.670 3,188.970
Note: p<0.1; p<0.05; p<0.01
Regression Results - India private colleges
Dependent variable:
n_colleges n_college_binary
OLS normal glm: quasipoisson probit
link = log
(1) (2) (3) (4)
dyn -0.277** -0.277** -0.154*** -0.070
(0.117) (0.117) (0.045) (0.072)
Year fixed effects Yes Yes Yes Yes
State fixed effects Yes Yes Yes Yes
Observations 4,631 4,631 4,631 4,631
R2 0.288
Adjusted R2 0.282
Akaike Inf. Crit. 23,555.670 3,188.970
Note: p<0.1; p<0.05; p<0.01

2.2.2 Model with controls - 2009 :2019

India colleges - Poisson model
fe2
Dependent Var.: n_colleges
dyn -0.1164 (0.0813)
incumbentTRUE -0.1962. (0.1135)
margin_percentage -0.0121** (0.0040)
turnout_percentage 0.0005 (0.0081)
enop 0.1071 (0.0742)
constituency_typeSC -0.2997. (0.1617)
constituency_typeST 0.3477. (0.1804)
log(electors) 0.2561 (0.7724)
no_terms -0.0412* (0.0180)
log(total_assets) 0.0123 (0.0197)
serious_crime -0.0075 (0.0399)
non_serious_crime 0.0126 (0.0237)
edu -0.0136. (0.0076)
Fixed-Effects: ——————
year_el Yes
State Yes
___________________ __________________
S.E.: Clustered by: State
Observations 2,763
Squared Cor. 0.35990
Pseudo R2 0.34152
BIC 11,599.0