Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
West Bengal
( With specific focus on urban areas )
Aishani Sengupta
Bidyutparna Mitra
Megha Das
Rajeeb Malik
Sasmit Dutta
Ujani Bhattacharya
May 2018
ACKNOWLEDGEMENT
We would like to take this opportunity to convey our sincere gratitude to our Project
Supervisor, Dr. Tanmoyee Banerjee , Department of Economics, Jadavpur University for her
unending guidance and support.
We would also like to express our regards and gratitude to Dr. Arpita Ghose (Dhar), Head of
the Department of Economics, Jadavpur University and all other professors of the
Department for their valuable inputs throughout the process and for organising a powerpoint
presentation in the month of November to check our progress. We are also extremely
thankful to the non-teaching staff of the Department, the Departmental Library of Economics
and the Students‟ Computer laboratory for lending a helping hand whenever we needed and
for providing us with research materials and well-equipped computers.
We are also grateful to all authors whose valuable and seminal works have guided us
throughout this research work.
Last but not the least, we would like to specially thank all our classmates for sharing their
opinions and views. We are extremely grateful to all our survey respondents who have
rendered us their valuable time.
Without the cooperation of the aforementioned, we could never have imagined transforming
our ideas into reality and completing this project in time.
1
Table of Contents
List of figures 4
1. Introduction 5
2. Methodology 10
3. Literature Review 15
4. Data Overview 18
2
5.1 Regression 1: To estimate the effect of Kanyashree grant on annual proportion of
5.2 Regression 2: To estimate the effect of Kanyashree grant on raising preferred age of
marriage 45
6. Conclusion 53
Appendix A 57
Appendix B 60
References 62
3
List of Figures
Fig. 4.1.1 Distribution of observations on the basis of receiving K1 grant ............................ 18
Fig. 4.2.1 Distribution of observations across three districts ................................................. 19
Fig. 4.2.2 Distribution of Kanyashree recipients across districts of Residence ..................... 20
Fig. 4.3 Distribution of observations based on poverty line.................................................... 21
Fig. 4.3.2 Proportion of recipients and non-recipients of Kanyashree among BPL card
holders...................................................................................................................................... 22
Fig. 4.3.3 Proportion of Kanyashree recipients and non-recipients among non-holders of
BPL cards................................................................................................................................. 22
Fig. 4.4.1 Division of students according to class of study ..................................................... 23
Fig. 4.4.2 Division of Kanyashree recipients based on class of study..................................... 24
Fig. 4.5.1 Division of students according to caste ................................................................... 25
Fig. 4.5.2 Division of Kanyashree recipients on the basis of caste ......................................... 25
Fig. 4.5.3 Division of Kanyashree Recipients on the basis of religion .................................... 26
Fig. 4.6.1 Division of families on the basis of own or rented house ........................................ 27
Fig. 4.6.2 Division of Kanyashree recipients on the basis of own or rented house ................ 28
Fig. 4.7.1 Educational qualification of mothers ...................................................................... 29
Fig. 4.7.2 Educational qualification of fathers ........................................................................ 30
Fig. 4.8.1 Frequency distribution of preferred age of marriage ............................................. 31
Fig.4.8.2 Preferred age of marriage - recipients vs. non-recipients ....................................... 32
Fig. 4.9 Proportion of girls with bank accounts - before and after receiving grant ............... 33
Fig. 4.9.1 Proportion of bank account holders among Kanyashree recipients ....................... 34
Fig. 4.9.2 Proportion of bank account holders among non-recipients .................................... 34
Fig. 4.10 Utilisation of Kanyashree grant ............................................................................... 35
Fig. 4.11 Plans on pursuing higher education beyond Higher Secondary Level (with grant vs.
without grant) ..................................................................................................................... 36
4
1. Introduction
Education is considered a basic human right. According to article 21A, of Indian constitution
the state shall provide free and compulsory education to all children of the age of six to
fourteen years. However, under patriarchal form of Indian society, girls are often denied their
right to education and in many cases, they are married off even before the age of eighteen.
Our state West Bengal is no exception. In West Bengal, it is observed that girls often dropout
of school mainly due to poverty, high education cost, and lack of proper infrastructure.
Under the Prohibition of Child Marriage Act, 2006, 18 is the legal age of marriage for girls,
and 21 for boys in India. Despite this act being in existence for several years, child marriage
has continued to be rampant in West Bengal. According to District Level Household Survey
3, 2007-08, the state ranked fifth highest in the country when it came to the prevalence of
child marriage. While the trend is more openly observed and common in rural areas, it was
observed than more than 25% of girls in non-slum urban areas also reported to getting
After the enactment of the PCMA 2006, The Department of Women Development and Social
Welfare and Child Development (DWCD), West Bengal, supported by UNICEF, took several
daughters making compliance to the law enforceable, but to no avail. Several factors worked
against the measures implemented. Primarily, the problem with eradicating child marriage
lies in deeply embedded social constructs about the role of men and women in a patriarchal
1
Rapid Assessment, Kanyashree Prakalpa (2015)
https://www.wbkanyashree.gov.in/readwrite/publications/000113.pdf
5
society, the value placed on the girl child - centuries old ideas that cannot be reversed by
According to Dutta and Sen (2016), evidences show that female education induces higher
levels of income and growth. Thus, a number of policy interventions have been introduced in
most developing countries including India, to augment enrolment and retention rates, while
bridging the gender gap at the same time by offering incentives linked to issues like
attendance. Early marriage is one of the important reasons behind female drop out at the
secondary level. Given that, West Bengal has the highest share of under-age marriage in
Kanyashree Prakalpa is an initiative taken by the Government of West Bengal to improve the
life and the status of the girls by helping economically backward families with cash support
so that families do not arrange the marriage of their girl child before eighteen years and
radical departure from the state‟s previous approach in child marriage prevention.
Kanyashree is a conditional cash transfer scheme which targets those most at-risk for child
marriage: adolescent girls between the ages of 13 and 18, and its conditions are aimed at
incentivizing their education and discouraging child marriage till at least age 18, the legal age
of marriage. The purpose of this initiative is to uplift those girls who are from poor families
and thus can‟t pursue higher studies due to tough economic conditions. It has been given
2
Kanyashree Prakalpa Rapid Assessment (2015)
https://www.wbkanyashree.gov.in/readwrite/publications/000113.pdf
6
and the UNICEF. Over 4 million adolescent girls have already enrolled in the scheme and
about $500 million handed over to the beneficiaries through their bank accounts.
1. Annual scholarship of Rs. 750 for unmarried girls aged 13-18 years enrolled in
2. One time grant of Rs. 25,000 for girls when they reach the age of 18, provided they
are pursuing educational, vocational or technical training courses at that time (K2).
The annual scholarship is for unmarried girls aged 13-18 years enrolled in class VII-XII in
courses.
Both the benefits under the scheme are granted to the girls who belongs to families with
annual family income not more than Rs. 1,20,000. The income bar is waived for the girls who
Justice Home. The scheme has covered 40 lakh school and college going girls under its
umbrella.
In our study of the literature concerning Kanyashree Prakalpa and other studies on CCTs and
analysis of impact of CCT schemes, we noticed that studies on Kanyashree Scheme were rare
and whatever work has been done thus far, fails to focus on addressing the changes in
expenditure patterns in households brought about by the CCT scheme. We were motivated to
study the impact this scheme was having on expenditure pattern of the households, especially
in the case of educational expenditure. Further, we wanted to study the effect this scheme was
having on catapulting girl children into higher education, and if the lump sum transfer in
7
Primary Objectives:
Our objective is to explore the overall effect of the cash transfer by dividing our study into
1. Study the effect of yearly cash transfer of Rs.750 (K1) on yearly household expenditure on
2. Study the effect of this scheme on the preferred age of marriage of the girl child, that is,
whether the scheme has induced in them the willingness to continue higher education and
Secondary Objectives:
Our study on this Conditional Cash Transfer scheme mainly focuses on how the policy has
been effective in delaying the marriageable age of girls in West Bengal and also its impact on
the educational expenditure. The effectiveness of the policy has been found out using
econometric analysis. The regression has been conducted for girls enrolled for the
Kanyashree Prakalpa scheme as well as for the ones who are not eligible for the grant.
8
However, after conducting the survey in three different districts of West Bengal (in and
around areas surrounding Kolkata), we expect to find that the yearly conditional cash transfer
of Rs.750 has a negligible effect on the yearly educational expenditure of the girl child. It is
mainly seen that that annual fees of the school they study in is much higher than the cash
On the other hand, the lump sum amount of Rs.25000 which they get after completing their
higher secondary examination serves as an incentive for them to complete their school
education and study further. They tend to focus on their career and continue their higher
studies which results in delaying their marriageable age. This is a positive effect of the
scheme as women become more independent and are capable of finding jobs instead of
The rest of the dissertation is organised in the following manner: In section 2 we have
presented the methodology of this project. Section 3 presents a brief survey on existing
literature on different types of conditional cash transfer programme. The section 4 gives an
exploration and descriptive analysis of the survey data. Section 5 presents the results of the
regression analysis. Finally, Section 6 concludes and presents possible policy prescription of
the project.
9
2. Methodology
We had sample points of cross-sectional data which included beneficiaries and non-
The survey has been conducted in the three districts : Kolkata, South 24 Parganas, and North
24 Parganas.
Initially, the survey was to be conducted using the technique of simple random sampling. This
is the process of selection of a group of units in such a manner that every unit has an „equal
chance‟ of being included in the sample. However, this was not a feasible technique to
conduct the survey due to time constraint, low response rate and lack of proper data available.
schools according to our own judgment and conducted the survey in the respective schools.
The purposive sample is basically a non-probability sample that is selected based on the
The method of carrying out the survey was to give out 30 questionnaires to the students of
class VIII, IX and X and 30 questionnaires to the students of class XI and XII. We selected
The data set includes samples points from the following five schools :
10
3.Dum Dum Baidyanath Institution for Girls, Dum Dum
The questionnaire has been designed to comprehend the effect of different factors on the
educational expenditure and change in preferred age of marriage of the girl child.
1. Name
2. Guardian‟s name
3. Birth date
4. Religion
5. Caste
6. Address
7. District
8. Class of study
9. Name of school
11
10. Medium of instruction in school
2. Educational qualification of the respondent‟s father, mother and her siblings(if any)
4. Whether or not the family owns a BPL (Below Poverty Line) Card
This section mainly checks whether the girl child is eligible for receiving the Kanyashree
grant or not. If the annual income of the family is less than 1,20,000, only then will she be
eligible to apply for the scheme. Additionally, this section was designed to explore whether
there is any correlation between families with BPL cards and receiving the Kanyashree grant.
Section C was designed specifically for recipients of the grant only and consists of questions
which help us study about the effect of the Kanyashree scheme on the girl child‟s educational
12
2. How the money is utilised - Whether it is used for her educational expenditure in the
present or saved for future use in the same, whether it is saved for the girl child‟s
3. If the total expenditure on education has increased, decreased or remained the same
4. Whether the girl child was engaged in any other profession before Kanyashree
5. Whether the girl child was engaged in any other profession before Kanyashree
7. Hours of self study and domestic work done by the girl child before the scheme
8. If the girl child has a bank account after availing the scheme ( this was specifically
included to check if the policy was having an effect towards increasing financial
10. Hours of self study and domestic work done by the girl child after receiving the grant
11. Preferred age of marriage of the girl child to assess whether she is more inclined
13. If the girl child is interested in continuing her studies after school
13
14. Whether the girl child is interested in joining college, getting married or working after
15. Whether she would have continued with her studies in the absence of the grant
16. Whether the grant should be increased, why it should be increased and by what
These questions mainly help us to find out whether the yearly grant of Rs.750 has any
positive effect on the educational expenditure of the girl child. Also, it gives us information if
the girl child is interested in pursuing higher education given that she will get a lump sum
14
3. Literature Review
We reviewed literature on two basic topics: Conditional Cash Transfers (CCTs) and specific
Our study of existing literature on Conditional Cash Transfer Programmes across the world
Based on a study of several CCTs aimed at empowering the girl child implemented in
developing countries across the world (some of these programmes include Apni Beti Apna
Dhan (India), Zomba Cash Transfer Program (Malawi), Female Secondary School
Assistance Programme (Bangladesh)), Amin et al, 2016 argue that providing CCTs is
effective in temporarily improving education and health outcomes, but changing underlying
norms about child marriage will require implementation of such CCT programmes along with
rights-based approach.
According to Lincove et al., 2016 targeting CCTs to certain groups of children can improve
efficiency, but positive effects on eligible children may be offset by reductions in investments
for ineligible siblings. Using data from Nicaragua, the authors estimate program effects on
eligible children and older siblings who aged out of eligibility. They find that CCTs had the
15
largest effects on eligible children, but older brothers also benefited. These results suggest
that income effects of CCTs apply to both eligible and ineligible children.
A study by Das Gupta et al., 2008 identified legislation and 58 policy and program
interventions to delay age at marriage in India from 1961 through 2005. The authors found
out that:
1. Education continues to be the single most important predictor of age at marriage over time
(this explains the motivation behind programmes such as Kanyashree Prakalpa which aim to
2. Most of these programmes failed to work in India because of a lack of a proper monitoring
system in place.
3. Echoing the argument by Amin et al (2016), this study too concluded that programmes
with the goal of delaying girls‟ age at marriage can only be successful by addressing
underlying social norms around sexuality and chastity, and reassuring parents about their
daughters‟ safety.
A study by Nanda et al., 2014 on the Apni Beti Apni Dhan (ABAD) CCT scheme
implemented in Haryana found that girls in ABAD program are more likely to stay in school
A study conducted by Field et al., 2016 on the different methods of delaying marriageable
age in Bangladesh found that financial incentives in the form of cash transfers are the most
effective.
Studies specifically on Kanyashree Prakalpa were rare. Two serious evaluations are
underway at the time of writing this report, one by Pratichi Institute and another by Dutta and
16
Sen (2016). Some literature that was available revealed to us the following facts about the
While incidences of child marriage is still prevalent in West Bengal, there is a positive
change in attitude among the community, parents and the beneficiary in terms of delaying the
age of marriage. Though the ill effect of early marriage was not known to many, because of
the One-time Grant, most of the responding parents indicated willingness to delay marrying
While most parents were inclined to utilising the money from the transfer for their daughter‟s
education, few of the parents also view Kanyashree as a support for their daughters‟
marriage.4
Sen and Dutta (2018) concluded that the scheme has had the following impacts:
3.The K2 transfer (Rs. 25,000 lump sum transfer) has been broadly used in the following
fields:
Given the various studies on the impact of conditional cash transfer, we try to find out the
effects of the cash transfer under Kanyashree Prakalpa on family education expenditure on
the girl child and the preferred marital age of the girls in the urban area.
3
Kanyashree Prakalpa, A Rapid Assessment (2014)
https://www.wbkanyashree.gov.in/readwrite/publications/000111.pdf
4
Baseline Survey for Socioeconomic Indicators in West Bengal
(2014)https://www.wbkanyashree.gov.in/readwrite/publications/000110.pdf
17
4. Data Overview
In this section we explore the entire dataset to study the trends and patterns observed.
Data from a total of 200 girl students of classes VIII, IX, IX, X1 and XII were collected, with
respect to studying the K1 grant. Approximately 49.5% of the students were beneficiaries of
the K1 grant. It was observed that a significant portion of the students of class VIII had not
attained the necessary age to receive the grant at the time of filling the questionnaire.
A total of 200 observations were collected from Kolkata district and neighbouring North 24
18
The distribution data points across these districts are illustrated below.
The highest number of data points were obtained from North 24 Parganas, followed by
We further divided the Kanyashree recipients and non-recipients according to their districts
19
Fig. 4.2.2 Distribution of Kanyashree recipients across districts of
Residence
The majority of Kanyashree recipients within our sample were reported to be from North 24
Parganas. But, this result could be due to the fact that majority of our observations were from
Dum Dum area in North 24 Parganas so that might have resulted in a bias. We performed an
ANOVA test5 to compare the mean family incomes of the subjects across three districts of
residence and the p-value was found to be insignificant. Therefore, we could not reject the
null hypothesis that mean income across all districts might be the same. So, the higher
5
See Appendix B for detailed results
20
4.3 Proportion of observations Below Poverty Line
Of the total observations, about 41.7% of the respondents reported that their families own a
BPL Card. From this, we concluded that about 41.7% of the respondents belong to families
21
Fig. 4.3.2 Proportion of recipients and non-recipients of Kanyashree
among BPL card holders
We further found out proportion of BPL card holders and non-holders who are recipients of
the grant or not. 48% of BPL card holders were found to be recipients of the grant, while 52%
22
Among respondents from families without a BPL card, it was found that the ratio of
Majority of the respondents were reported to be from class IX, followed by classes X and XI.
Students from class XII could not be interviewed due to the time of survey clashing with
For recipients of the grant, majority of recipients were reported to be in class X and XI .
A very small proportion of students in class VIII reported to being beneficiaries, primarily
because most of them have not attained the required age for availing the grant.
23
Fig. 4.4.2 Division of Kanyashree recipients based on class of study
Of all observations, majority were found to be of General unreserved class (Fig. 4.5.1).
Of the recipients of the grant, again majority were observed to be General (Fig. 4.5.2).
This result does not indicate that on an average recipients are of unreserved caste. It merely
reflects the fact that majority of the sample points included belonged to unreserved castes.
24
Fig. 4.5.1 Division of students according to caste
Based on religion, majority of student in our survey were found to be Hindus. This could be
because the schools that we chose to survey were located in primarily Hindu areas.
25
From all observations, 197 were found to be Hindus, 2 Christians and only 1 respondent was
Muslim.
From among recipients of grants, Hindus were the majority too, because of the bias in our
data resulting from the choice of areas that we covered. This data thus is not fully
representative of the correlation between religion and reach of the grant in the overall
population.
26
4.6 Division of families based on own or rented house
We had asked this question to determine whether the diversion of a part of income towards
rent was having any effect on the expenditure on education of the girl-child.
27
Fig. 4.6.2 Division of Kanyashree recipients on the basis of own or
rented house
Majority of mothers of respondents were reported to have an education level of upto class 10.
The same trend was seen among fathers too, majority of whom reported to have completed
education upto class 10 level. However, the proportion of those with education from classes 6
to 10 level was larger among mothers than fathers. A larger proportion of fathers are seen to
have completed education beyond class X or even class XII compared to mothers.
28
Fig. 4.7.1 Educational qualification of mothers
29
Fig. 4.7.2 Educational qualification of fathers
Respondents were required to report their preferred age of marriage, and majority of them
30
Fig. 4.8.1 Frequency distribution of preferred age of marriage
We also studied mean preferred age of marriage between recipients and non-recipients of the
grant.
For students that don‟t benefit from K1, mean preferred age of marriage was found to be 24.
For students that benefit from K1, mean preferred age of marriage was found to be approx.
26.
From the graph, we note that willingness to marry at later ages (29-31) is much higher among
beneficiaries of K1 grant.
31
Fig.4.8.2 Preferred age of marriage - recipients vs. non-recipients
Respondents who are beneficiaries of the grant were asked to report whether they owned a
bank account prior to receiving the grant. Based on this information, we calculated the
proportion of the students who had bank accounts irrespective of the grant and the proportion
of students who have bank accounts after receiving the grant. We see that the proportion
increases.
We observe that proportion of students with bank accounts has risen, after receiving the
grant. So we could say that as a secondary effect, Kanyashree Prakalpa has acted as a
proponent for bringing a larger portion of female students under the banking system.
32
Fig. 4.9 Proportion of girls with bank accounts - before and after
receiving grant
Non-recipients of the grant were also asked to report if they own a bank account. Proportion
of girls with bank accounts were found to be higher among recipients of the grant compared
33
Fig. 4.9.1 Proportion of bank account holders among Kanyashree
recipients
34
Clearly, proportion of bank account holders among recipients stands at 93.8% while among
67.7% of girl students who were K1 beneficiaries used the funds for expenditure on
32.3% of the beneficiaries planned on saving the funds for future expenditure on education.
None of the students reported that the K1 grant was being used for domestic expenditure or
6
It is important to note that ideally, 100% bank account coverage should have been expected from the sample
involving recipients. However, our study failed to address the question of why the figure stands at 93.8%. We
urge future researchers on this topic to investigate possible factors that can be attributed to this outcome.
35
4.11 Effect of K1 grant on planning of pursuing higher education after H.S
From the sample set, it was observed that 128 students want to pursue higher education after
However 16 students reported that they would pursue higher education if and only if they
Thus, there is a significant 12.5% increase in willingness to pursue higher education solely
due to K1 grant.
36
5. Models and Regression Results
To explore the impact of Kanyashree Prakalpa, we divided our study into two parts :
(i) studying the effect of Kanyashree grant on educational expenditure on the girl child
(ii) to study whether the grant was having any effect on raising the preferred age of marriage
among students.
Accordingly, we formulated two models and ran regressions to estimate the results.
that arises in cross-sectional data, we resorted to reporting our results with Generalized Least
Squares Regression.
StateSE 12 has been used to conduct the econometric analysis of the data.
Sections 5.1 and 5.2 will elaborate on the control variables, model specifications, models
estimated from regression analysis and detailed analysis of the results obtained.
7
See Appendix B
37
5.1 Regression 1: To estimate the effect of Kanyashree grant on annual
Of the 200 data points we had collected, 24 had to be dropped due to erroneous reporting of
expenditure details. We dropped those observations for which people did not report their
expenditure details or in cases where the proportion of expenditure spent on the education of
We hypothesized that the proportion of annual educational expenditure on the girl child 8 (a
ratio of total yearly educational expenditure on the girl child and total consumption
2. District in which the girl child lives in North 24 Parganas, South 24 Parganas or Kolkata.
This has 2 variables- dis_south and dis_north . A dummy has been taken for each:
8
Note: Annual educational expenditure includes annual school tuition fees, cost of school uniform, expenditure
on books and stationery items, private tuition fees, travel cost, expenditure on extracurricular activities and other
miscellaneous unanticipated expenses required for educational purposes.
38
= 0 , otherwise
= 0 , otherwise
We included this variable to check the impact of place of residence on the educational
the assumption that a large family size may adversely affect the educational expenditure of
the girl child as the family has to spend more on consumption expenditure.
expenditure on girl child for the general category households vis-à-vis reserved category
households.
The house ownership status has been included as a proxy for asset level of the households.
39
The yearly proportion of educational expenditure to total expenditure is taken as the
Null Hypothesis:
We hypothesize that the yearly Rs. 750 transfer will not have a significant effect on the
Model Description:
variables. It shows the marginal effect of those variables on the proportion of educational
40
Regression results:
Number of observations
Source SS df MS
= 176
0.23905555
Model 1.43433331 6 F(6,169) = 13.42
1
0.01780797
Residual 3.00954719 169 Prob>F = 0.0000
2
0.02539360
Total 4.4438805 175 R-squared = 0.3228
3
Adj R-squared = 0.2987
Root MSE = 0.13345
Table 5.1.1 shows the OLS results of regression run on cross-sectional data, and we find that
We see p-value is significant, so we discard null hypothesis of homoscedastic data and accept
9
See Appendix B
41
To smoothen out the heteroskedasticity in the data, we resort to estimating coefficients using
GLS regression.
Number of observations =
Linear Regression
176
F(6,169) = 11.02
Prob>F = 0.0000
R-squared = 0.3228
Root MSE = 0.13345
Robust
eduexo_propo
Coeff.. Std. Err. t p>|t| 95% Confidence Interval
rtiopn
Table 5.1.2 gives us the results of the regression run to estimate the effect of Kanyashree
1. This result shows that house_status , dis_north and totalnoofmembers are significant
as the p-value of these variables are 0.004,0.000 and 0.017 respectively, which are
less than 0.05 at 95% confidence interval. As the p value of these three variables are
less than 0.05, we reject the null hypothesis and conclude that they have a significant
42
effect on the proportional educational expenditure of the girl child.
This also implies that eduexpo_proportion is higher for students residing in Kolkata,
the reference category, than the students in in North 24 parganas . However the
in Kolkata may be attributed to the high tuition fees, transport cost compared to 24
Parganas(North).
that household which have to pay rent have to divert a portion of their income on rent,
Again, the fact that we can see a negative significant relation between proportion of
expenditure on education of girl child and total number of family members is logical.
The greater the number of family members, lesser is the expenditure expected to be
0.705, 0.758 and 0.810 respectively, that is, greater than 0.05 at 95% confidence
interval which means that we accept the null hypothesis and conclude that these
expenditure of the girl child. Thus as expected, we see that the annual Kanyashree
grant of Rs.750 has no significant effect on the girl child‟s educational expenditure
independent varibles in the model. In this case the model explains 32.28% of the
variance in eduexo_propor~n .
43
4. Prob>F is the p-value of the model. It indicates the reliability of X (independent
than 0.05 which shows that there is a statistical relationship between them.
5. Root MSE is the root mean squared error which is the standard deviation of the
regression. The closer to zero, the better the fit. In this model, the value of root MSE
10
See Appendix B
44
5.2 Regression 2: To estimate the effect of Kanyashree grant on raising
One of the goals in this project is to explore the effect which the cash grant given through
Kanyashree has on the general age of marriage associated with teenage girls. We had
collected data on the preferred age of marriage of girls and on the average age of marriage of
females in the family. It was observed that the age preferred by the respondents mostly did
not match with the average family trend. Hence the age difference was computed between the
two and a regression was run in order to study how it varied with different variables.
For this regression, we could not consider all 200 data points due to erroneous data obtained.
Some respondents had not reported their preferred age of marriage, so we had to exclude
these data points from the regression. Further data points had to be dropped from the
regression since average age of marriage of females in the family was not reported. Finally,
64 such observations were found and discarded and ultimately the regression was run using
Control Variables:
1. Caste
45
1. Caste : The dummy for caste named as “caste_dummy” is generated as follows -
= 1, otherwise
= 0 , otherwise
= 0 , otherwise
generated as follows -
= 0 , otherwise
4. House ownership status : The dummy for this variable named as “house_status” was
generated as follows -
5. Kanyashree recipient status : The dummy for this variable namely “receive_kanyashree”
46
receive_kanyahsree = 1 , if the respondent is a recipient of Kanyashree
= 0 , otherwise
Now the variable “agediff” was generated as agediff = (preferred age of marriage - average
The variable “agediff” was then taken as the dependent variable and the regression was run
caste_dummy
dis_north
dis_south
house_status
mother_work
47
Model Specification:
β5 house_status + β6 mother_work
Regression results:
Number of observations =
Source SS Df MS
136
Model 112.063036 6 18.6771727 F(6,169) = 2.76
Residual 873.047258 129 6.7678082 Prob>F = 0.0148
Total 985.110294 135 7.29711329 R-squared = 0.1138
Adj R-squared = 0.0725
Root MSE = 2.6015
receive_kanya
1.608997 0.4591145 3.5 0.001 0.7006273 2.517366
shree
caste_dummy 0.3985348 0.4827362 0.83 0.411 -0.5565707 1.35364
dis_north 0.359457 0.5188739 0.69 0.49 -0.6671477 1.386062
dis_south -0.4275953 0.6341784 -0.67 0.501 -1.682333 0.8271422
mother_work -0.1952617 0.6957514 -0.28 0.779 -1.571823 1.181299
house_status -0.120429 0.623071 -0.19 0.847 -1.35319 1.112332
_cons 0.9972361 0.7167905 1.39 0.167 -0.4209515 2.415424
48
However, we again conduct a heteroskedasticity check for the data. 11
Again we see that p-value is significant, so we can conclude that heteroskedasticity is present
in the data.
Number of observations =
Linear Regression
136
F(6,169) = 3.10
Prob>F = 0.0071
R-squared = 0.1138
Root MSE = 2.6015
Robust
receive_kanyas
1.608997 0.4632727 3.47 0.001 0.6924004 2.525593
hree
caste_dummy 0.3985348 0.516259 0.77 0.442 -0.6228963 1.419966
dis_north 0.359457 0.5480959 0.66 0.513 -0.7249642 1.443878
dis_south -0.4275953 0.4751989 -0.9 0.27 -1.367788 0.5125973
mother_work -0.1952617 0.6215815 -0.31 0.754 -1.425076 1.034552
house_status -0.120429 0.7396617 -0.16 0.871 -1.583868 1.34301
_cons 0.9972361 0.7029546 1.42 0.158 -0.3935768 2.388049
11
See Appendix B
49
Conclusions drawn from Regression 2:
(< 0.01 , hence we reject the null that the coefficient = 0). It has a positive coefficient of
1.608997 implying that the age difference between the preferred age of marriage and the
average age of marriage of girls in the family varies positively with the girl being a recipient
This result is instrumental in showing that the Kanyashree grant is indeed having a positive
effect on inducing girl children to look forward to a future in which getting married at a very
2. All other variables are insignificant in their effect on the dependent variable, with P-
values ranging from 0.158 to 0.871, all values being > 0.05.
3. The R-squared value = 0.1138 implying that 11.38% of the model‟s variations being
T-tests:
1. In addition to the regression, we also conducted a paired t-test or a mean test between
the preferred age of marriage of respondents and average age of marriage of females
in the family. The null hypothesis states that the mean difference between preferred
age of marriage and average age of marriage in the family for the girls receiving
After conducting the paired t test, we see that the p value is 0.000 which is less than 0.05 at
95% confidence interval. Thus, we reject the null hypothesis and conclude that the mean
50
difference between preferred and average age of marriage for the girls receiving the
The results we have obtained are based on a very small sample size, which isn‟t very well-
representative of the overall population of students within that income group and age group
all over West Bengal. A larger and more representative sample would have yielded better
results.
A major issue we faced while conducting the research is underreporting of income and
Only five schools were considered for the study, from primarily Hindu neighbourhoods,
which restricted us from seeing the variations of effect of Kanyashree, caused on the basis of
religion.
The questionnaire did not address other factors which could have been used as controls in the
The variations probably could have been explained better by approximating nonlinear
models, but nonlinear regression models are beyond the scope of our undergraduate syllabus
12
See Appendix B for results
51
Most of these issues arose due to paucity of time and low response rate. A larger scale of
study, spanning all districts of West Bengal, more representative of the population of the
state, incorporating more control variables would have yielded much better results.
52
6. Conclusion
We have been able to gather some very interesting results in our study on the effects of
A primary criticism aimed at cash transfer programmes is that the money will be used for
purposes other than the policy requirements of the programme. However, the annual transfer
component K1 isn‟t exactly supposed to serve the purpose of an income supplement (its
purpose has been elaborated on in the following paragraphs). Irrespective of this, respondents
reported to either spending the money for current educational expenditure or saving it for
inclusion of underprivileged young females in the society. The incentives of the grant are also
clearly seen to be working towards raising mean preferred age of marriage among recipients,
thereby fulfilling one of its primary policy goals. This shows that the grant, understandably
the lump sum amount of Rs. 25,000, is opening up new avenues for young girls with limited
means. This is also reflected in our finding of how many of the girls who would not have
opted to continue with higher education in the absence of the grant are willing to do so now.
However, whether this is happening as a result of the K1 grant or the K2 grant is yet to be
We do however study the effect of the K1 annual grant of Rs. 750 on the proportion of annual
consumption expenditure spent on education of girl child, in order to assess if cash was being
used in the present or saved for the future. Our study revealed that the annual transfer has not
53
been significant in affecting annual educational expenditure on the girl child. It could be
because the Rs. 750 annual grant is supposed to work as a bait mechanism to push the girl-
child towards completing her education up to higher secondary level, after which she is to
receive a Rs. 25,000 lump sum amount. Respondents also reported that the annual transfer
had made no change in the annual educational expenditure, which is in tune with our
findings. Thus, we can conclude that the annual transfer‟s purpose is more to act as a bait to
child. However, in a country with a male-child bias, to uplift an entire generation of female
children out of a poverty trap and raise their standard of living, more importance has to be
placed on the role of the cash transfer as a supplement to educational expenditure on the
female child. The amount of the transfer should be able to at least cover the basic minimum
of expenses on school fees, school uniform, books and stationery and other basic needs. On
this basis, the transfer amount should be increased to at least Rs. 2000 per annum. Otherwise,
in the face of rapid rising educational expenditures, parents will ultimately find it more
profitable to marry off their daughters than continue with their education. In the same vein,
the lump sum transfer of Rs. 25,000 could either be increased to a larger amount or the grant
should be accompanied with some complementary vocational training course which would
ultimately help the girl child enter the workforce and become self-sufficient, instead of
simply getting a degree up to the higher secondary level, which, in today‟s job market is of
little value. The idea should be to endow the beneficiary with a particular set of skills to enter
54
One of our findings is that 100% banking coverage was not observed in case of recipients of
the grant. This observation is completely out of tune with what is the usual expectation since
Another issue we think is important is that many respondents who qualify for the grant under
age and family income criterion had still not been enrolled under the scheme. Negligence on
behalf of the school administration was cited by many such respondents. This issue too needs
to be resolved.
Cash transfer schemes have always been seen as populist schemes, and this question arises
even in the case of Kanyashree Prakalpa. Principals in two of the schools reported that many
families were underreporting their income by forging documents, in order to avail the grant
(primarily for the Rs. 25,000 lump sum transfer). It was pointed out that such cases were
being willfully ignored by authorities, since one beneficiary meant acquiring the trust and
votes of the entire family. This issue is of grave concern and needs immediate attention. Care
should be taken so that this potentially successful scheme does not become an instrument of
politicking.
We had planned to explore whether Kanyashree was helping girl-children abandon side-jobs
and pursue education full time. However, in our sample we did not find any student who held
a job before receiving the grant. Hence, this issue remains unexplored but we think it is a
55
Our limited scope of study left us with very few instruments to analyse this issue in further
detail. This leaves the door open for future work to be conducted on this programme. Since it
is a relatively new programme (Kanyashree Prakalpa has been in existence since 2013),
concrete data, especially those on income and expenditure patterns of families are still scarce.
In a few years‟ time, we expect to find better data on income and expenditure patterns of
households, and the effect of the transfer on career decisions made by beneficiaries of the
grant.
56
Appendix A
Appendix A provides detailed summary of descriptive statistics of all control variables and
dependent variables used in the regression analysis in sections 5.1 and 5.2 of this paper.
Figures A.5.1.1, A.5.1.2 and A.5.1.3 provide summary statistics of dependent, independent
57
Table A.5.1.3 Summary Statistics of Control Variables of Regression 1
Figures A.5.2.1, A.5..2 and A.5.2.3 provide summary statistics of dependent, independent and
58
Table A.5.2.3 Summary Statistics of Control Variables of Regression 2
59
Appendix B
Residence
Table B.4.2.1 Summary Statistics and ANOVA test results for variation in income across 3 districts
Regression 1
60
Table B.5.1.2 Correlation between Annual Income and status of house ownership
13
In the above table, “var 2” is average age of marriage of females in the family and “var 3” is preferred age of
marriage of respondent.
61
References
Ready Lessons from India on Delaying Marriage for Girls. ICRW Publications
Dutta, A. and Sen, A. (2016). Kanyashree Prakalpa in West Bengal Justification and
prakalpa-in-west-bengal-india-justification-and-evaluation
Programme Needs More Push From State and Beneficiaries. EPW Engage. Vol. 53, Issue
Project
eligible children and their siblings. Education Economics, Volume 24, Issue 4
62
Baseline Survey of Socio-Economic Indicators in West Bengal (2014)
https://www.wbkanyashree.gov.in/readwrite/publications/000100.pdf
https://www.wbkanyashree.gov.in/readwrite/notice_publications/kp_baseline_survey.pdf
https://www.wbkanyashree.gov.in/readwrite/publications/000111.pdf
https://www.wbkanyashree.gov.in/readwrite/publications/000113.pdf
63
64