SlideShare a Scribd company logo
1 |C h u
Justin Chu
11/9/15
ECON 171a
India’s Sanitation Push: Impact on Mortality
Introduction:
For the past two decades, the Indian government has made a strong push to eliminate
open defecation, especially in rural areas due to the prevalence of death through rotaviruses –
viruses transmitted through the fecal-oral route. These viruses cause nearly 334,000 deaths
among children out of the 2.3 million deaths in India each year – nearly 15% of all deaths.1
The
main cause for such prevalence of these diseases in India is a lack of sanitary toilets in many
village centers which causes many residents to resort to open defecation. Starting in 1992, the
Indian government launched a program called Nirmal Bharat Abhiyan (renamed the Total
Sanitation Campaign in 2012) to incentivize community-led total sanitation (CLTS). This CLTS
effort revolved around providing localized means of construction through toilet construction
subsidies and lump sums to villages marked as “open defecation free.”2
. After his election in
2014, Indian Prime Minister Narendra Modi launched Swachh Bharat Abhiyan (Clean India
Mission) which launched the biggest cleanliness drive in Indian history.3
The majority of this
program focuses on the construction of Individual Household Latrines (further known as
IHHLs), cluster, community, and school toilets in all Gram Panchayats (municipalities).4
Part of
this program brought the monitoring into the social media realm with the inception of an app
used to take pictures and track progress of each finished construction project in a central internet
database. This database is sorted by time, state, and construction type; by sorting this data with
1
“Rotavirus mortality in India: estimates based on a nationally representative survey of diarrhoeal deaths,” World
Health Organization, Bulletin of the World Health Organization, 2012; vol. 90: 720-727.
2
“Restructuring of the Nirmal Bharat Abhiyan into Swachh Bharat Mission,” Press Information Bureau,
Government of India, Sep 24 2014.
3
Ibid.
4
Ibid.
2 |C h u
diagnosed deaths from diarrheal diseases, one can begin to regress the effects of the new toilet
implementation program on overall health.
Literature Review:
For the bulk of the data in terms of application of the program, the source will be the
SBM website which carries the completed construction data based on the number of unique
geotags placed by each district user. Unfortunately, access does not always equate to utilization
as many of the rural poor feel as though these new toilets are not beneficial or necessary to their
own lives.5
In addition, many of our human development numbers on an annual basis will come
from the Center Bureau for Health Intelligence as well as the Annual Status of Education Report
which is facilitated by Pratham. As was previously stated, these numbers may not truly reflect
the overall impact of these individual variables in India; however, they offer very comprehensive
health and education surveys respectively, especially on the state-level. By utilizing these more
focused studies, one can create a more satisfactory model.
The main area of focus for the research will be the states of Uttar Pradesh, Madhya
Pradesh, and Rajasthan which are all along the northern edge of India. These states were chosen
due to their BIMARU status which is an acronym of the state names which is coincidentally – or
not – very similar to the Hindi word bimar meaning "sick." Bihar was not chosen due to the lack
of reported data in terms of both the SBM/NBA and disease death related data. These BIMARU
states share geographical boundaries as well as the fact that they all lag behind the rest of India in
terms of overall development factors such as real GDP growth and human development goals.
By targeting these less-developed areas, the impact of the program itself as opposed to other state
5
“PM Modi fulfills promise of 80 lakh toilets but not many takers in rural India,” Sharma, Neetu: India Today, Aug
14, 2015. Web.
3 |C h u
factors should be more distinct especially given their relatively similar low strata of each state
involved.
In terms of other studies referenced, press releases and other similar nongovernmental
organizations reflecting general conditions of water and sanitation within the borders; however,
the bulk of the literature will come from the data provided by the state level and the reflections
provided in the model. In addition, the Indonesian government started a similar program which
can serve part of understanding the challenges facing India. Since the program is still developing,
many of the studies and accompanying surveys that would address the SBM’s efficacy are not
completed; despite that, India, and especially the Indian government, utilized similar structures
in their push to mitigate air and water pollution which is analyzed in a paper by Michael
Greenstone and Rema Hanna called Environmental Regulations, Air and Water Pollution, and
Infant Mortality in India. While the aforementioned government programs are not perfect
representations of the SBM, they each serve as important proxy lessons to create an analytic
apparatus for the SBM. The model combined with third-party research and on-the-ground stories
in newspapers and magazines will serve as the basis of the policy conclusions by combining
empirical statistics and insider perspective to create the most accurate picture as many of the
numbers can be deceptive without utilizing the most pertinent information.
Model
The basis of the model is a non-linear state-fixed effects regression in which nearly all of
the variables are being measured with the natural logarithm transformation applied to them. The
natural logarithmic transformation on the dependent variable (Yjt), total instances of diarrhea or
diarrheal deaths, and the independent variables: total toilets constructed, GDP per capita, total
Model 1 - Nonlinear regression using income
4 |C h u
sanitation facilities (for drinking water) constructed, boils down the analysis to the forms of
percentages. For example, a 10% increase of an independent variable will result in a 1.10*β (the
coefficient in front of the dependent variable) change in the dependent variable. By using a non-
linear approach, the implicit assumption is that toilet construction, while holding all other factors
constant, does not have a fixed impact on instances of disease and death regardless of existing
facilities. In addition, the state fixed effects are measured in the (n-1) model in which the first
state is dropped and measured through the constant and the other states are attributed to certain
state ids.
In the above equation, β1 represents overall toilets constructed. β2 is measured in level 1 –
a baseline test for literacy issued to children which serves as a proxy for parental education since
many of these illiterate children most likely come from illiterate households and vice versa –
rural literacy rates. β3 is the GDP per capita as measured by overall GDP produced by each state
divided by the population and measured in crores, and β4 is the annual state production of
covered water dispensaries to combat clean water problems in the area. In addition, µt represents
the fixed effects of each state (Uttar Pradesh, Madhya Pradesh, and Rajasthan). As is the case in
all regressions, these variables were chosen to address any omitted variables that may have an
impact on Yjt beyond toilet construction. Literacy and extended education as a result, income,
and construction of clean water projects all could have an impact on conception of an easily-
transmitted, yet easily-avoided malady like acute diarrhea. The observations take five years’
worth of data on the state-by-state annual statistics; therefore, with the five years from 2010-
2014 and the three states, there are 15 observations in the panel data model.
However, the model can be tweaked in a more focused way to understand how the SBM
and NBA affected people and consequently – the impact of the program on disease. For the
5 |C h u
NBA, the program paid households subsidies to build the toilets in their own homes. The
government offered 4600 rupees to any homeowner, especially poor home owners, to construct
their own individual toilets.6
As a result, there exists an interaction factor between individual
toilet constructions because building a toilet has the cash incentive which may make the
homeowner more likely to begin construction of the toilet, consequently, impacting their disease
instances. Due to the multicollinearity of GDP per capita and the interaction term of the
continuous variable of GDP per capita and toilets constructed, the variable for GDP per capita is
dropped out and exchanged for the interaction term.
Model 2 - New regression with interaction term
Data
For the 15 panel observations, most of the data falls into a relatively reasonable range
when applying the natural logarithmic transformation; unfortunately, when trying to expand the
observations by increasing the time scale from annual to quarterly or monthly is not applicable as
many of the observations fluctuate in random patterns as the construction numbers are delayed in
their reporting. Subsequently, the same problem emerges when attempting to use district level
data as the other data such as cases of acute diarrhea or literacy rate are not accurately reported
for each year. Therefore, using the annual data is the only way to insure validity for the
measurements within the data. Unfortunately, when attempting to run the extended fixed effects
regression model, the increased number of independent variables minimized the observations in
favor of degrees of freedom used in the F-test which is a measure to illustrate the fit of a
regression model on the data.
6
“Restructuring of the Nirmal Bharat Abhiyan into Swachh Bharat Mission,” Press Information Bureau,
Government of India, Sep 24 2014.
6 |C h u
Looking at Figure 5, while the absolute totals of the variables do suffer from
occasionally large variances and, consequently, high standard deviations, they still form only a
small fraction of the mean in most cases meaning the points do not differ too much as a
proportion of the mean. That point can be illustrated in examining the kurtosis of the tails which
indicate the amount of outliers within the observations; most of these points do not exceed the
“baseline” value of 3 implying the tails are not oversaturated with observations which would
hamper the OLS assumptions of the model. For the total values, especially for the construction
numbers, one must expect a rather large fluctuation between the points given the differences
between states which will be covered more in the model results.
Looking at Figure 6, one can see that the dependent variable (“lntotalcases”) does not
have a very large standard deviation which is indicative of the logarithmic transformation as the
natural logarithm requires rather large values to fluctuate in value; however, it is also indicative
that the percentages of those contracting these symptoms do not vary greatly from year to year.
The same can be said for the independent variables (“lntoilets”, “lnBPLtoilets”, “lnGDPPC”,
“lev1rurallit, lnsharespent”). In addition, the kurtosis and skewness of each variable except
“lnsharespent” reflects a relatively normal distribution of the points indicates there are no strong
outliers which show strength in the data.
Case Studies
The main analytic lens applied for overall success for these types of programs in a similar
situation will be the implementation of the toilet construction initiative in Indonesia; the crucial
paper is a working paper written by the World Bank entitled Impact Evaluation of a Large-Scale
Rural Sanitation Project in Indonesia written by Lisa Cameron, Manisha Shah, and Susan Olivia
(hereafter referred to as Cameron et al.) The writers studied eight rural districts in 160
7 |C h u
communities in Indonesia attempting to study the Total Sanitation and Sanitation Market
(TSSM) project.7
While the toilet construction in the treated communities increased by three
percentage points and diarrhea prevalence was only 3.3 percent in the treated areas compared to
4.6 in those that were not, it did not impact sanitation habits such as washing hands, open
defecation, or using soap which also have a large impact on diarrhea conception.8
When
measuring certain baseline demographic descriptive statistics such as defecting in an improved
sanitation facility, in the open, washing hands after going to the toilet, and having soap, none of
the respective difference-in-difference p-values are statistically significant except for the
construction of toilet variable, for obvious reasons.9
Unfortunately, that means that the program’s
efficacy was not explicitly tied to toilet construction but the decreases in diarrhea in the control
area may be correlated to other factors outside of the scope of the TSSM. For the SBM, this
factor is important in attempting to discern the impact of toilets and knowledge of good
sanitation practices by itself.
The supplementary case study is the piece written by Greenstone and Hanna in which
they discuss the efficacy of the Indian government attempting to rein in air and water pollution:
the Supreme Court Action Plans and Mandated Catalytic Converters, as well as the National
River Conservation Plan.10
Through a Quandt likelihood ratio test, the researchers found that
there was almost no impact from the adoption of the National River Conservation Plan (NRCP)
but a definitive structural break from the air pollution policies from which very changes emerged
as an engaged public galvanized the additional push needed to execute the policy despite less
7
“Impact Evaluation of a Large-Scale Rural Sanitation Project in Indonesia,” Cameron et.al, The World Bank,
Water and Sanitation Program, Feb 2013.
8
Ibid.
9
Ibid.
10
“Environmental Regulations, Air and Water Pollution, and Infant Mortality in India,” Greenstone et al,
Massachusetts Institute of Technology, Feb 2011.
8 |C h u
than efficient government structures.11
The researchers concluded that this break emerged from a
factor of utility juxtaposed with more energized and efficient institutions (the Supreme Court)
that could effectively legislate and dole out proper regulations and create enforcement
mechanisms.12
In addition, the study showed that the efficacy of the program was tied to the
high avoidance costs for air pollution as people can only breathe air in a certain medium and
cannot buy imported air while water is more easily cleaned through boiling it.13
The main
conclusion from these points to the eventual success or failure or the SBM is that the government
policies can succeed when the populace stands behind the program and encourages their own
part of the policy. Another conclusion is that the program functions best when the avoidance
costs are high and people feel a definitive change in their avoidance of the pollutants, especially
in a comparison between water and air pollutants.
Results
While much of the program is rather new, the results gathered were a mixed affair in
terms of expectations. Before the experimentation with the model, many of the stories regarding
the efficacy of the SBM stated that many of those who had toilets constructed in their homes or
local townships felt too constrained to use the new toilets provided to them and still chose to
defecate in the open. Due to these untraceable biases, many of the independent variables lack the
statistical significance needed to assert their own validity as variables. While it is possible that
the lack of significance for some of these variables is due to the overall lack of observations,
looking at the graphs in Figure 2 clearly shows trends that would support the present hypothesis
that increased facilities and increased pressure to use them will decrease the instances of diarrhea
11
“Environmental Regulations, Air and Water Pollution, and Infant Mortality in India,” Greenstone et al,
Massachusetts Institute of Technology, Feb 2011.
12
Ibid.
13
Ibid.
9 |C h u
in these poor states. For example, the regression results in Figure 3 using the robust standard
error assumption still show the same relative t-statistic which implies that these observations are
relatively consistent even if the statistical significance is rather small compared to the desired
level. However, with that being said, there is clearly some form of bias that is accounting for the
fluctuations in the total instances which is clearly encapsulated in the variables marking the state
id in both Figure 1 and Figure 3 as state id 2 (Madhya Pradesh) and state id 3 (Rajasthan) are
statistically significant to a reasonable degree. In addition, the variable for GDP per capita
measured a positive variable, meaning that increases in overall GDP per capita may be
detrimental to health; however, one must consider the impact of the subsidies on the toilet and
see if that extra incentive at all impacts the overall levels of diarrhea.
While looking at the other model comparing the interaction factor of GDP per capita and
total below poverty line toilets constructed in a logarithmic transformation in Figure 4, there is a
clear significance level of both the construction of toilets below the poverty line and the subsidy
put in place for that construction. Unfortunately, the signs for both variables are negative and
positive respectively meaning that the impact of the subsidy for each toilet built is negatively
increasing or time. Thus, the model implies that the impact of toilet building with the subsidy,
which is already built into the legislation, the toilets themselves are only effective to a certain
point in a case of diminishing margins of returns. While many of the independent variables in
this regression lack overall statistical significance, it is important to include them as they serve to
eliminate any omitted variable bias. Whether or not the overall independent variables are
statistically significant, the individual state levels must also be considered in order to gain the
full picture.
10 |C h u
In analyzing both Model 1 and Model 2, there is a clear statistical significance in both
Madhya Pradesh (state 2) and Rajasthan (state 3) as opposed to Uttar Pradesh (constant).
Additionally, the signs on both of the states are negative which indicates that there are
underlying factors within each state that are not necessarily outlined in the model which help
lower the percentages of diarrhea occurrence. Therefore, there is some unknown factor or factors
in the model which shows that the states Madhya Pradesh and Rajasthan are healthier than Uttar
Pradesh beyond the variables stated in the two different regressions.
Conclusions and Policy
The program push for a “Clean India” is certainly not without its warts or inefficiencies;
however, the impact of toilets and toilet usage is still yet to be determined under the SBM and
NBA. Unfortunately, it is clear that toilets are not the only factor in overall health due to the fact
that they are not necessarily the biggest variable factor in the models; this news is of no real
surprise given the lessons provided in Indonesia with their TSSM program. While looking at the
stories surrounding the program itself, the reception has not been incredibly outstanding. In an
interview with Kartikay Mehrotra of Bloomberg Business, a woman named Sunita in Uttar
Pradesh stated that “only Dalits, the lowest Hindu caste, should be exposed to excrement in a
closed space.”14
That mentality is the main problem in the overall longevity of the program;
without people being willing to use the toilets, they are effectively useless in preventing
instances of diseases. As was the case in Indonesia, one can lead the program and build all the
sanitary conditions for toilets as necessary but they are not useful when left idle. As stated by
Archana Patkar, a program manager at the Water Supply & Sanitation Collaborative Council, “‘
the problem has gotten worse with the government thinking this is a supply driven problem. The
14
“India’s Toilet Race Failing as Villages Don’t Use Them,” Mehrotra, Kartikay, Bloomberg Business, Aug 4,
2014. Web.
11 |C h u
problem is that germs are invisible, and so understanding the threat of open defecation is far
removed from reality.”15
Therefore, the next step is doing internal research on the mentality
surrounding toilet usage to enlighten people to the changes that clean defecation practices can
have in their lives. Without properly addressing the mentality around the usage of toilets as
something only meant for lower strata of society, there is only so much that can be accomplished
by building toilets without any education. For example, when the World Bank released their
study on Indonesia, surveys on good sanitary behavior were released to both the treatment and
control group on their understanding of these topics; these results showed that the changes in
construction of sanitary areas did not create a significant impact on beliefs around proper
sanitation.16
By educating people below the poverty line, as the caste is not necessarily tied to
economic status, there can be a real impact of using those toilets on diarrhea death and diseases;
however, as the old saying goes, you can lead a horse to a toilet but you cannot make it use the
toilet.
15
“India’s Toilet Race Failing as Villages Don’t Use Them,” Mehrotra, Kartikay, Bloomberg Business, Aug 4,
2014. Web.
16
“Impact Evaluation of a Large-Scale Rural Sanitation Project in Indonesia,” Cameron et.al, The World Bank,
Water and Sanitation Program, Feb 2013.
12 |C h u
Figures:
Figure 1 – fixed effects regression using below poverty line household toilet construction
measurements.
Figure 2 - Graphs showing the changes in total cases of diarrhea and other indicator
variables measured by state id (1- UP, 2 - Madhya Pradesh, 3- Rajasthan)
_cons 3.836265 6.993411 0.55 0.598 -12.29057 19.9631
_Istate_id_3 -1.325917 .5417518 -2.45 0.040 -2.575199 -.0766347
_Istate_id_2 -.4617604 .4158162 -1.11 0.299 -1.420634 .4971135
lnGDPPC 1.230642 .6481366 1.90 0.094 -.2639639 2.725248
lev1rurallit .0206392 .0965196 0.21 0.836 -.2019354 .2432137
lntotalsancov -.1230762 .090466 -1.36 0.211 -.3316911 .0855387
lnBPLtoilets -.19465 .1377828 -1.41 0.195 -.5123778 .1230778
lntotalcases Coef. Std. Err. t P>|t| [95% Conf. Interval]
Robust
Root MSE = .27938
R-squared = 0.7461
Prob > F = 0.0002
F(6, 8) = 21.29
Linear regression Number of obs = 15
i.state_id _Istate_id_1-3 (naturally coded; _Istate_id_1 omitted)
. xi: regress lntotalcases lnBPLtoilets lntotalsancov lev1rurallit lnGDPPC i.state_id, r
13 |C h u
Figure 3 - Same regression as Figure 1 but with robust heteroskedastic errors
Figure 4- Same regression as Figures 1 and 3 but including a subsidy interaction term
(which drops out GDP variable)
_cons 1.40949 6.554861 0.22 0.835 -13.70605 16.52503
_Istate_id_3 -1.439068 .49245 -2.92 0.019 -2.57466 -.3034764
_Istate_id_2 -.5625891 .3866586 -1.46 0.184 -1.454226 .3290473
lnsharespent -.1044713 .0528974 -1.97 0.084 -.226453 .0175104
lnGDPPC 1.310905 .5904148 2.22 0.057 -.0505935 2.672404
lev1rurallit .0103106 .098293 0.10 0.919 -.2163535 .2369747
lnBPLtoilets -.1013052 .1005108 -1.01 0.343 -.3330836 .1304732
lntotalcases Coef. Std. Err. t P>|t| [95% Conf. Interval]
Robust
Root MSE = .26847
R-squared = 0.7656
Prob > F = 0.0004
F(6, 8) = 16.82
Linear regression Number of obs = 15
i.state_id _Istate_id_1-3 (naturally coded; _Istate_id_1 omitted)
. xi: regress lntotalcases lnBPLtoilets lev1rurallit lnGDPPC lnsharespent i.state_id, r
_cons 3.836247 6.993453 0.55 0.598 -12.29068 19.96318
_Istate_id_3 -1.325916 .5417537 -2.45 0.040 -2.575203 -.0766302
_Istate_id_2 -.461761 .4158177 -1.11 0.299 -1.420638 .4971164
lnGDPxBPLtoilets 1.230642 .6481394 1.90 0.094 -.26397 2.725254
lev1rurallit .0206393 .09652 0.21 0.836 -.2019362 .2432148
lntotalsancov -.1230756 .0904663 -1.36 0.211 -.3316912 .08554
lnBPLtoilets -1.425292 .7257946 -1.96 0.085 -3.098977 .2483936
lntotalcases Coef. Std. Err. t P>|t| [95% Conf. Interval]
Robust
Root MSE = .27938
R-squared = 0.7461
Prob > F = 0.0002
F(6, 8) = 21.29
Linear regression Number of obs = 15
i.state_id _Istate_id_1-3 (naturally coded; _Istate_id_1 omitted)
> ate_id, r
. xi: regress lntotalcases lnBPLtoilets lntotalsancov lev1rurallit lnGDPxBPLtoilets i.st
14 |C h u
99% 660.3888 660.3888 Kurtosis 5.810532
95% 660.3888 373.6622 Skewness 1.712458
90% 373.6622 257.7996 Variance 26962.84
75% 256.8474 256.8474
Largest Std. Dev. 164.2036
50% 144.026 Mean 183.3494
25% 56.7074 56.7074 Sum of Wgt. 15
10% 43.5264 54.2441 Obs 15
5% 3.75 43.5264
1% 3.75 3.75
Percentiles Smallest
sharespent
99% 72391.2 72391.2 Kurtosis 2.136755
95% 72391.2 65658.42 Skewness .5419822
90% 65658.42 61480.77 Variance 1.94e+08
75% 58636.92 58636.92
Largest Std. Dev. 13943.92
50% 42496.59 Mean 45256.68
25% 33366.44 33366.44 Sum of Wgt. 15
10% 30038.24 32087.83 Obs 15
5% 26539.93 30038.24
1% 26539.93 26539.93
Percentiles Smallest
GDPPC
99% 2915407 2915407 Kurtosis 6.888059
95% 2915407 1613384 Skewness 2.03895
90% 1613384 1166016 Variance 4.74e+11
75% 900769 900769
Largest Std. Dev. 688177.2
50% 648792 Mean 818640.1
25% 515427 515427 Sum of Wgt. 15
10% 252800 266197 Obs 15
5% 134873 252800
1% 134873 134873
Percentiles Smallest
totaltoilets
99% 1042578 1042578 Kurtosis 4.097271
95% 1042578 711103 Skewness 1.377213
90% 711103 621743 Variance 7.53e+10
75% 472521 472521
Largest Std. Dev. 274358.4
50% 213312 Mean 328844.5
25% 158175 158175 Sum of Wgt. 15
10% 81700 102905 Obs 15
5% 45359 81700
1% 45359 45359
Percentiles Smallest
totalBPLtoilets
99% 826246 826246 Kurtosis 2.096739
95% 826246 768021 Skewness -.2490236
90% 768021 745457 Variance 3.66e+10
75% 740328 740328
Largest Std. Dev. 191327.5
50% 535012 Mean 539293.5
25% 431893 431893 Sum of Wgt. 15
10% 227571 290705 Obs 15
5% 223106 227571
1% 223106 223106
Percentiles Smallest
totalcases
Figure 5- Summary Statistics of all dependent and independent variables in absolute form
15 |C h u
99% 6.492829 6.492829 Kurtosis 5.626488
95% 6.492829 5.923352 Skewness -1.469123
90% 5.923352 5.552183 Variance 1.445991
75% 5.548482 5.548482
Largest Std. Dev. 1.202494
50% 4.969994 Mean 4.76409
25% 4.037905 4.037905 Sum of Wgt. 15
10% 3.773368 3.993494 Obs 15
5% 1.321756 3.773368
1% 1.321756 1.321756
Percentiles Smallest
lnsharespent
99% 16.2 16.2 Kurtosis 2.161826
95% 16.2 15.9 Skewness .2154488
90% 15.9 15.1 Variance 2.265524
75% 15.1 15.1
Largest Std. Dev. 1.505166
50% 13.3 Mean 13.58667
25% 12.7 12.7 Sum of Wgt. 15
10% 11.4 12.4 Obs 15
5% 11.2 11.4
1% 11.2 11.2
Percentiles Smallest
lev1rurallit
99% 11.18984 11.18984 Kurtosis 1.938605
95% 11.18984 11.09222 Skewness .1607812
90% 11.09222 11.02648 Variance .0917857
75% 10.97912 10.97912
Largest Std. Dev. .3029615
50% 10.65718 Mean 10.67694
25% 10.41531 10.41531 Sum of Wgt. 15
10% 10.31023 10.37623 Obs 15
5% 10.18641 10.31023
1% 10.18641 10.18641
Percentiles Smallest
lnGDPPC
99% 13.85721 13.85721 Kurtosis 2.60935
95% 13.85721 13.47457 Skewness -.1402715
90% 13.47457 13.34028 Variance .700332
75% 13.06584 13.06584
Largest Std. Dev. .8368584
50% 12.27051 Mean 12.39531
25% 11.97146 11.97146 Sum of Wgt. 15
10% 11.31081 11.54156 Obs 15
5% 10.72236 11.31081
1% 10.72236 10.72236
Percentiles Smallest
lnBPLtoilets
99% 13.62465 13.62465 Kurtosis 2.629621
95% 13.62465 13.55157 Skewness -.8406101
90% 13.55157 13.52175 Variance .1756777
75% 13.51485 13.51485
Largest Std. Dev. .4191393
50% 13.19004 Mean 13.1254
25% 12.97593 12.97593 Sum of Wgt. 15
10% 12.33522 12.58006 Obs 15
5% 12.3154 12.33522
1% 12.3154 12.3154
Percentiles Smallest
lntotalcases
Figure 6- Summary Statistics of all dependent and independent variables in natural logarithm
form
16 |C h u
References:
1. “ASER Yearly Reports”, Aser Centre, Pratham Education Foundation,
http://asercentre.org, Web.
2. “CHBI National Health Profiles”, Central Bureau of Health Intelligence,
http://cbhidghs.nic.in , Web.
3. “Environmental Regulations, Air and Water Pollution, and Infant Mortality in India,”
Greenstone et al, Massachusetts Institute of Technology, Feb 2011.
4. “Impact Evaluation of a Large-Scale Rural Sanitation Project in Indonesia,” Cameron
et.al, The World Bank, Water and Sanitation Program, Feb 2013.
5. “India’s Toilet Race Failing as Villages Don’t Use Them,” Mehrotra, Kartikay,
Bloomberg Business, Aug 4, 2014. Web.
6. “India’s Toilet Race Failing as Villages Don’t Use Them,” Mehrotra, Kartikay, Bloomberg
Business, Aug 4, 2014. Web.
7. “National Rural Drinking Water Programme”, Government of India,
http://indiawater.gov.in , Web.
8. “PM Modi fulfills promise of 80 lakh toilets but not many takers in rural India,”
Sharma, Neetu: India Today, Aug 14, 2015. Web.
9. “Restructuring of the Nirmal Bharat Abhiyan into Swachh Bharat Mission,” Press
Information Bureau, Government of India, Sep 24 2014.
10. “Swachh Bharat Mission”, Government of India, http://tsc.gov.in , Web.

More Related Content

What's hot

MRC HIVAN Forum 25 October 2011
MRC HIVAN Forum 25 October 2011MRC HIVAN Forum 25 October 2011
MRC HIVAN Forum 25 October 2011
info4africa
 
The impact of the global financial crisis on reproductive and maternal health...
The impact of the global financial crisis on reproductive and maternal health...The impact of the global financial crisis on reproductive and maternal health...
The impact of the global financial crisis on reproductive and maternal health...
UN Global Pulse
 
Trends in future health financing and coverage: future health spending and un...
Trends in future health financing and coverage: future health spending and un...Trends in future health financing and coverage: future health spending and un...
Trends in future health financing and coverage: future health spending and un...
Henar Rebollo Rodrigo
 
Supplementary Actuarial Analysis of HIV/AIDS in Lagos State, Nigeria
Supplementary Actuarial Analysis of HIV/AIDS in Lagos State, NigeriaSupplementary Actuarial Analysis of HIV/AIDS in Lagos State, Nigeria
Supplementary Actuarial Analysis of HIV/AIDS in Lagos State, Nigeria
HFG Project
 
Lagos state hiv response
Lagos state hiv responseLagos state hiv response
Lagos state hiv response
NigerianBusinessCoal
 
Minnesota’s accountable communities for health
Minnesota’s accountable communities for health Minnesota’s accountable communities for health
Minnesota’s accountable communities for health
soder145
 
Impact and cost effectivene of rotavirus vaccine introduction in afghanistan
Impact and cost   effectivene of  rotavirus vaccine  introduction in afghanistanImpact and cost   effectivene of  rotavirus vaccine  introduction in afghanistan
Impact and cost effectivene of rotavirus vaccine introduction in afghanistan
Najibullah Safi
 
Kazal 6e data sources, uses and limitations
Kazal 6e   data sources, uses and limitationsKazal 6e   data sources, uses and limitations
Kazal 6e data sources, uses and limitations
Sizwan Ahammed
 
Minnesota Accountable Health Model Continuum of Accountability Assessment: Ev...
Minnesota Accountable Health Model Continuum of Accountability Assessment: Ev...Minnesota Accountable Health Model Continuum of Accountability Assessment: Ev...
Minnesota Accountable Health Model Continuum of Accountability Assessment: Ev...
soder145
 
Kano State Health Profile - Nigeria
Kano State Health Profile - NigeriaKano State Health Profile - Nigeria
Kano State Health Profile - Nigeria
HFG Project
 
Wed vs ondcp labelle
Wed vs ondcp labelleWed vs ondcp labelle
Wed vs ondcp labelle
OPUNITE
 
Implementation of bphsphc afghanistan experience august 2019
Implementation of bphsphc afghanistan experience august 2019Implementation of bphsphc afghanistan experience august 2019
Implementation of bphsphc afghanistan experience august 2019
Najibullah Safi
 
IOSR Journal of Pharmacy (IOSRPHR), www.iosrphr.org, call for paper, research...
IOSR Journal of Pharmacy (IOSRPHR), www.iosrphr.org, call for paper, research...IOSR Journal of Pharmacy (IOSRPHR), www.iosrphr.org, call for paper, research...
IOSR Journal of Pharmacy (IOSRPHR), www.iosrphr.org, call for paper, research...
iosrphr_editor
 
District Layyah Budget Trends Analysis for Advocacy by LR 19042015
District Layyah Budget Trends Analysis for Advocacy by LR 19042015District Layyah Budget Trends Analysis for Advocacy by LR 19042015
District Layyah Budget Trends Analysis for Advocacy by LR 19042015
DUNYA NEWS
 
Post reform changes in health care access and affordability in MN
Post reform changes in health care access and affordability in MN Post reform changes in health care access and affordability in MN
Post reform changes in health care access and affordability in MN
soder145
 
Family Planning impact brief-Bangladesh
Family Planning impact brief-BangladeshFamily Planning impact brief-Bangladesh
Family Planning impact brief-Bangladesh
Golam Kibria MadhurZa
 
New perspectives on global healthspending UHC
New perspectives on global healthspending UHCNew perspectives on global healthspending UHC
New perspectives on global healthspending UHC
Chuchai Sornchumni
 
00.SM2015-Technical Note July 2015- ENG
00.SM2015-Technical Note  July 2015- ENG00.SM2015-Technical Note  July 2015- ENG
00.SM2015-Technical Note July 2015- ENG
Emma Margarita Iriarte Carcamo
 
Effective implementation of national health strategy final
Effective implementation of national health strategy finalEffective implementation of national health strategy final
Effective implementation of national health strategy final
Najibullah Safi
 
Health information
Health information Health information
Health information
Dr.Kuntala Ray
 

What's hot (20)

MRC HIVAN Forum 25 October 2011
MRC HIVAN Forum 25 October 2011MRC HIVAN Forum 25 October 2011
MRC HIVAN Forum 25 October 2011
 
The impact of the global financial crisis on reproductive and maternal health...
The impact of the global financial crisis on reproductive and maternal health...The impact of the global financial crisis on reproductive and maternal health...
The impact of the global financial crisis on reproductive and maternal health...
 
Trends in future health financing and coverage: future health spending and un...
Trends in future health financing and coverage: future health spending and un...Trends in future health financing and coverage: future health spending and un...
Trends in future health financing and coverage: future health spending and un...
 
Supplementary Actuarial Analysis of HIV/AIDS in Lagos State, Nigeria
Supplementary Actuarial Analysis of HIV/AIDS in Lagos State, NigeriaSupplementary Actuarial Analysis of HIV/AIDS in Lagos State, Nigeria
Supplementary Actuarial Analysis of HIV/AIDS in Lagos State, Nigeria
 
Lagos state hiv response
Lagos state hiv responseLagos state hiv response
Lagos state hiv response
 
Minnesota’s accountable communities for health
Minnesota’s accountable communities for health Minnesota’s accountable communities for health
Minnesota’s accountable communities for health
 
Impact and cost effectivene of rotavirus vaccine introduction in afghanistan
Impact and cost   effectivene of  rotavirus vaccine  introduction in afghanistanImpact and cost   effectivene of  rotavirus vaccine  introduction in afghanistan
Impact and cost effectivene of rotavirus vaccine introduction in afghanistan
 
Kazal 6e data sources, uses and limitations
Kazal 6e   data sources, uses and limitationsKazal 6e   data sources, uses and limitations
Kazal 6e data sources, uses and limitations
 
Minnesota Accountable Health Model Continuum of Accountability Assessment: Ev...
Minnesota Accountable Health Model Continuum of Accountability Assessment: Ev...Minnesota Accountable Health Model Continuum of Accountability Assessment: Ev...
Minnesota Accountable Health Model Continuum of Accountability Assessment: Ev...
 
Kano State Health Profile - Nigeria
Kano State Health Profile - NigeriaKano State Health Profile - Nigeria
Kano State Health Profile - Nigeria
 
Wed vs ondcp labelle
Wed vs ondcp labelleWed vs ondcp labelle
Wed vs ondcp labelle
 
Implementation of bphsphc afghanistan experience august 2019
Implementation of bphsphc afghanistan experience august 2019Implementation of bphsphc afghanistan experience august 2019
Implementation of bphsphc afghanistan experience august 2019
 
IOSR Journal of Pharmacy (IOSRPHR), www.iosrphr.org, call for paper, research...
IOSR Journal of Pharmacy (IOSRPHR), www.iosrphr.org, call for paper, research...IOSR Journal of Pharmacy (IOSRPHR), www.iosrphr.org, call for paper, research...
IOSR Journal of Pharmacy (IOSRPHR), www.iosrphr.org, call for paper, research...
 
District Layyah Budget Trends Analysis for Advocacy by LR 19042015
District Layyah Budget Trends Analysis for Advocacy by LR 19042015District Layyah Budget Trends Analysis for Advocacy by LR 19042015
District Layyah Budget Trends Analysis for Advocacy by LR 19042015
 
Post reform changes in health care access and affordability in MN
Post reform changes in health care access and affordability in MN Post reform changes in health care access and affordability in MN
Post reform changes in health care access and affordability in MN
 
Family Planning impact brief-Bangladesh
Family Planning impact brief-BangladeshFamily Planning impact brief-Bangladesh
Family Planning impact brief-Bangladesh
 
New perspectives on global healthspending UHC
New perspectives on global healthspending UHCNew perspectives on global healthspending UHC
New perspectives on global healthspending UHC
 
00.SM2015-Technical Note July 2015- ENG
00.SM2015-Technical Note  July 2015- ENG00.SM2015-Technical Note  July 2015- ENG
00.SM2015-Technical Note July 2015- ENG
 
Effective implementation of national health strategy final
Effective implementation of national health strategy finalEffective implementation of national health strategy final
Effective implementation of national health strategy final
 
Health information
Health information Health information
Health information
 

Viewers also liked

Rubricadic
RubricadicRubricadic
WitlonPM_Broch Final Edit
WitlonPM_Broch Final EditWitlonPM_Broch Final Edit
WitlonPM_Broch Final Edit
Nick Murer
 
Niños y violencia una realidad compleja
Niños y violencia una realidad complejaNiños y violencia una realidad compleja
Niños y violencia una realidad compleja
Itzel Barrón Escovar
 
Freight and Transport Ground Services
Freight and Transport Ground ServicesFreight and Transport Ground Services
Freight and Transport Ground Services
Priyankawebspy
 
Lehendakariaren hitzalida - Aranzadi Gogora Institutua
Lehendakariaren hitzalida - Aranzadi Gogora InstitutuaLehendakariaren hitzalida - Aranzadi Gogora Institutua
Lehendakariaren hitzalida - Aranzadi Gogora Institutua
Irekia - EJGV
 
Pundit @Internet festival 2016
Pundit @Internet festival 2016Pundit @Internet festival 2016
Pundit @Internet festival 2016
Net7 srl
 
Vivaldi udberr
Vivaldi udberrVivaldi udberr
Vivaldi udberr
MUSIKAGALLEGOGORRIA
 
space solar power satellite
space solar power satellitespace solar power satellite
space solar power satellite
amit kumar
 

Viewers also liked (8)

Rubricadic
RubricadicRubricadic
Rubricadic
 
WitlonPM_Broch Final Edit
WitlonPM_Broch Final EditWitlonPM_Broch Final Edit
WitlonPM_Broch Final Edit
 
Niños y violencia una realidad compleja
Niños y violencia una realidad complejaNiños y violencia una realidad compleja
Niños y violencia una realidad compleja
 
Freight and Transport Ground Services
Freight and Transport Ground ServicesFreight and Transport Ground Services
Freight and Transport Ground Services
 
Lehendakariaren hitzalida - Aranzadi Gogora Institutua
Lehendakariaren hitzalida - Aranzadi Gogora InstitutuaLehendakariaren hitzalida - Aranzadi Gogora Institutua
Lehendakariaren hitzalida - Aranzadi Gogora Institutua
 
Pundit @Internet festival 2016
Pundit @Internet festival 2016Pundit @Internet festival 2016
Pundit @Internet festival 2016
 
Vivaldi udberr
Vivaldi udberrVivaldi udberr
Vivaldi udberr
 
space solar power satellite
space solar power satellitespace solar power satellite
space solar power satellite
 

Similar to Research Paper

Swachch bharat
Swachch bharatSwachch bharat
Swachch bharat
OpenSpace
 
Mec-105 2015-16 Free Ignou Assignment
Mec-105 2015-16 Free Ignou AssignmentMec-105 2015-16 Free Ignou Assignment
Mec-105 2015-16 Free Ignou Assignment
Rahul Singh
 
SocialCops and UN Papua New Guinea: Presentation for Data Stocktaking Workshop
SocialCops and UN Papua New Guinea: Presentation for Data Stocktaking WorkshopSocialCops and UN Papua New Guinea: Presentation for Data Stocktaking Workshop
SocialCops and UN Papua New Guinea: Presentation for Data Stocktaking Workshop
SocialCops
 
Bad Effects of Urbanization and Lifestyles, Population Health Improvements us...
Bad Effects of Urbanization and Lifestyles, Population Health Improvements us...Bad Effects of Urbanization and Lifestyles, Population Health Improvements us...
Bad Effects of Urbanization and Lifestyles, Population Health Improvements us...
IRJET Journal
 
Presentation1
Presentation1Presentation1
Presentation1
Sharjana Tehjib
 
Abebe et al-2020-systematic_reviews
Abebe et al-2020-systematic_reviewsAbebe et al-2020-systematic_reviews
Abebe et al-2020-systematic_reviews
Thomas Ayalew
 
Bardach final
Bardach finalBardach final
Bardach final
Aachal Devi
 
10 page Summary of TrackFin's results in Brazil
10 page Summary of TrackFin's results in Brazil10 page Summary of TrackFin's results in Brazil
10 page Summary of TrackFin's results in Brazil
TrackFin
 
An Empirical Study on the Change of Consumption Level of Chinese Residents
An Empirical Study on the Change of Consumption Level of Chinese ResidentsAn Empirical Study on the Change of Consumption Level of Chinese Residents
An Empirical Study on the Change of Consumption Level of Chinese Residents
Dr. Amarjeet Singh
 
Methodology Documentation
Methodology DocumentationMethodology Documentation
Methodology Documentation
Penggalih Herlambang
 
CBA Framework
CBA FrameworkCBA Framework
CBA Framework
blueboffin
 
Follow the Money: Making the Most of Limited Health Resources
Follow the Money: Making the Most of Limited Health ResourcesFollow the Money: Making the Most of Limited Health Resources
Follow the Money: Making the Most of Limited Health Resources
HFG Project
 
Follow the Money: Making the Most of Limited Health Resources
Follow the Money: Making the Most of Limited Health ResourcesFollow the Money: Making the Most of Limited Health Resources
Follow the Money: Making the Most of Limited Health Resources
HFG Project
 
Summary results of TrackFin's testing in Brazil, Ghana and Morocco
Summary results of TrackFin's testing in Brazil, Ghana and MoroccoSummary results of TrackFin's testing in Brazil, Ghana and Morocco
Summary results of TrackFin's testing in Brazil, Ghana and Morocco
TrackFin
 
MDR Roll out plan
MDR Roll out planMDR Roll out plan
MDR Roll out plan
dpmo123
 
Effect of Fiscal Independence and Local Revenue Against Human Development Index
Effect of Fiscal Independence and Local Revenue Against Human Development IndexEffect of Fiscal Independence and Local Revenue Against Human Development Index
Effect of Fiscal Independence and Local Revenue Against Human Development Index
Universitas Pembangunan Panca Budi
 
Nuhm
NuhmNuhm
Publications 2015 Final
Publications 2015 FinalPublications 2015 Final
Factors Affecting Consumer Health Care Services Delivery in Private Health Fa...
Factors Affecting Consumer Health Care Services Delivery in Private Health Fa...Factors Affecting Consumer Health Care Services Delivery in Private Health Fa...
Factors Affecting Consumer Health Care Services Delivery in Private Health Fa...
AI Publications
 
Phronesis
PhronesisPhronesis

Similar to Research Paper (20)

Swachch bharat
Swachch bharatSwachch bharat
Swachch bharat
 
Mec-105 2015-16 Free Ignou Assignment
Mec-105 2015-16 Free Ignou AssignmentMec-105 2015-16 Free Ignou Assignment
Mec-105 2015-16 Free Ignou Assignment
 
SocialCops and UN Papua New Guinea: Presentation for Data Stocktaking Workshop
SocialCops and UN Papua New Guinea: Presentation for Data Stocktaking WorkshopSocialCops and UN Papua New Guinea: Presentation for Data Stocktaking Workshop
SocialCops and UN Papua New Guinea: Presentation for Data Stocktaking Workshop
 
Bad Effects of Urbanization and Lifestyles, Population Health Improvements us...
Bad Effects of Urbanization and Lifestyles, Population Health Improvements us...Bad Effects of Urbanization and Lifestyles, Population Health Improvements us...
Bad Effects of Urbanization and Lifestyles, Population Health Improvements us...
 
Presentation1
Presentation1Presentation1
Presentation1
 
Abebe et al-2020-systematic_reviews
Abebe et al-2020-systematic_reviewsAbebe et al-2020-systematic_reviews
Abebe et al-2020-systematic_reviews
 
Bardach final
Bardach finalBardach final
Bardach final
 
10 page Summary of TrackFin's results in Brazil
10 page Summary of TrackFin's results in Brazil10 page Summary of TrackFin's results in Brazil
10 page Summary of TrackFin's results in Brazil
 
An Empirical Study on the Change of Consumption Level of Chinese Residents
An Empirical Study on the Change of Consumption Level of Chinese ResidentsAn Empirical Study on the Change of Consumption Level of Chinese Residents
An Empirical Study on the Change of Consumption Level of Chinese Residents
 
Methodology Documentation
Methodology DocumentationMethodology Documentation
Methodology Documentation
 
CBA Framework
CBA FrameworkCBA Framework
CBA Framework
 
Follow the Money: Making the Most of Limited Health Resources
Follow the Money: Making the Most of Limited Health ResourcesFollow the Money: Making the Most of Limited Health Resources
Follow the Money: Making the Most of Limited Health Resources
 
Follow the Money: Making the Most of Limited Health Resources
Follow the Money: Making the Most of Limited Health ResourcesFollow the Money: Making the Most of Limited Health Resources
Follow the Money: Making the Most of Limited Health Resources
 
Summary results of TrackFin's testing in Brazil, Ghana and Morocco
Summary results of TrackFin's testing in Brazil, Ghana and MoroccoSummary results of TrackFin's testing in Brazil, Ghana and Morocco
Summary results of TrackFin's testing in Brazil, Ghana and Morocco
 
MDR Roll out plan
MDR Roll out planMDR Roll out plan
MDR Roll out plan
 
Effect of Fiscal Independence and Local Revenue Against Human Development Index
Effect of Fiscal Independence and Local Revenue Against Human Development IndexEffect of Fiscal Independence and Local Revenue Against Human Development Index
Effect of Fiscal Independence and Local Revenue Against Human Development Index
 
Nuhm
NuhmNuhm
Nuhm
 
Publications 2015 Final
Publications 2015 FinalPublications 2015 Final
Publications 2015 Final
 
Factors Affecting Consumer Health Care Services Delivery in Private Health Fa...
Factors Affecting Consumer Health Care Services Delivery in Private Health Fa...Factors Affecting Consumer Health Care Services Delivery in Private Health Fa...
Factors Affecting Consumer Health Care Services Delivery in Private Health Fa...
 
Phronesis
PhronesisPhronesis
Phronesis
 

Research Paper

  • 1. 1 |C h u Justin Chu 11/9/15 ECON 171a India’s Sanitation Push: Impact on Mortality Introduction: For the past two decades, the Indian government has made a strong push to eliminate open defecation, especially in rural areas due to the prevalence of death through rotaviruses – viruses transmitted through the fecal-oral route. These viruses cause nearly 334,000 deaths among children out of the 2.3 million deaths in India each year – nearly 15% of all deaths.1 The main cause for such prevalence of these diseases in India is a lack of sanitary toilets in many village centers which causes many residents to resort to open defecation. Starting in 1992, the Indian government launched a program called Nirmal Bharat Abhiyan (renamed the Total Sanitation Campaign in 2012) to incentivize community-led total sanitation (CLTS). This CLTS effort revolved around providing localized means of construction through toilet construction subsidies and lump sums to villages marked as “open defecation free.”2 . After his election in 2014, Indian Prime Minister Narendra Modi launched Swachh Bharat Abhiyan (Clean India Mission) which launched the biggest cleanliness drive in Indian history.3 The majority of this program focuses on the construction of Individual Household Latrines (further known as IHHLs), cluster, community, and school toilets in all Gram Panchayats (municipalities).4 Part of this program brought the monitoring into the social media realm with the inception of an app used to take pictures and track progress of each finished construction project in a central internet database. This database is sorted by time, state, and construction type; by sorting this data with 1 “Rotavirus mortality in India: estimates based on a nationally representative survey of diarrhoeal deaths,” World Health Organization, Bulletin of the World Health Organization, 2012; vol. 90: 720-727. 2 “Restructuring of the Nirmal Bharat Abhiyan into Swachh Bharat Mission,” Press Information Bureau, Government of India, Sep 24 2014. 3 Ibid. 4 Ibid.
  • 2. 2 |C h u diagnosed deaths from diarrheal diseases, one can begin to regress the effects of the new toilet implementation program on overall health. Literature Review: For the bulk of the data in terms of application of the program, the source will be the SBM website which carries the completed construction data based on the number of unique geotags placed by each district user. Unfortunately, access does not always equate to utilization as many of the rural poor feel as though these new toilets are not beneficial or necessary to their own lives.5 In addition, many of our human development numbers on an annual basis will come from the Center Bureau for Health Intelligence as well as the Annual Status of Education Report which is facilitated by Pratham. As was previously stated, these numbers may not truly reflect the overall impact of these individual variables in India; however, they offer very comprehensive health and education surveys respectively, especially on the state-level. By utilizing these more focused studies, one can create a more satisfactory model. The main area of focus for the research will be the states of Uttar Pradesh, Madhya Pradesh, and Rajasthan which are all along the northern edge of India. These states were chosen due to their BIMARU status which is an acronym of the state names which is coincidentally – or not – very similar to the Hindi word bimar meaning "sick." Bihar was not chosen due to the lack of reported data in terms of both the SBM/NBA and disease death related data. These BIMARU states share geographical boundaries as well as the fact that they all lag behind the rest of India in terms of overall development factors such as real GDP growth and human development goals. By targeting these less-developed areas, the impact of the program itself as opposed to other state 5 “PM Modi fulfills promise of 80 lakh toilets but not many takers in rural India,” Sharma, Neetu: India Today, Aug 14, 2015. Web.
  • 3. 3 |C h u factors should be more distinct especially given their relatively similar low strata of each state involved. In terms of other studies referenced, press releases and other similar nongovernmental organizations reflecting general conditions of water and sanitation within the borders; however, the bulk of the literature will come from the data provided by the state level and the reflections provided in the model. In addition, the Indonesian government started a similar program which can serve part of understanding the challenges facing India. Since the program is still developing, many of the studies and accompanying surveys that would address the SBM’s efficacy are not completed; despite that, India, and especially the Indian government, utilized similar structures in their push to mitigate air and water pollution which is analyzed in a paper by Michael Greenstone and Rema Hanna called Environmental Regulations, Air and Water Pollution, and Infant Mortality in India. While the aforementioned government programs are not perfect representations of the SBM, they each serve as important proxy lessons to create an analytic apparatus for the SBM. The model combined with third-party research and on-the-ground stories in newspapers and magazines will serve as the basis of the policy conclusions by combining empirical statistics and insider perspective to create the most accurate picture as many of the numbers can be deceptive without utilizing the most pertinent information. Model The basis of the model is a non-linear state-fixed effects regression in which nearly all of the variables are being measured with the natural logarithm transformation applied to them. The natural logarithmic transformation on the dependent variable (Yjt), total instances of diarrhea or diarrheal deaths, and the independent variables: total toilets constructed, GDP per capita, total Model 1 - Nonlinear regression using income
  • 4. 4 |C h u sanitation facilities (for drinking water) constructed, boils down the analysis to the forms of percentages. For example, a 10% increase of an independent variable will result in a 1.10*β (the coefficient in front of the dependent variable) change in the dependent variable. By using a non- linear approach, the implicit assumption is that toilet construction, while holding all other factors constant, does not have a fixed impact on instances of disease and death regardless of existing facilities. In addition, the state fixed effects are measured in the (n-1) model in which the first state is dropped and measured through the constant and the other states are attributed to certain state ids. In the above equation, β1 represents overall toilets constructed. β2 is measured in level 1 – a baseline test for literacy issued to children which serves as a proxy for parental education since many of these illiterate children most likely come from illiterate households and vice versa – rural literacy rates. β3 is the GDP per capita as measured by overall GDP produced by each state divided by the population and measured in crores, and β4 is the annual state production of covered water dispensaries to combat clean water problems in the area. In addition, µt represents the fixed effects of each state (Uttar Pradesh, Madhya Pradesh, and Rajasthan). As is the case in all regressions, these variables were chosen to address any omitted variables that may have an impact on Yjt beyond toilet construction. Literacy and extended education as a result, income, and construction of clean water projects all could have an impact on conception of an easily- transmitted, yet easily-avoided malady like acute diarrhea. The observations take five years’ worth of data on the state-by-state annual statistics; therefore, with the five years from 2010- 2014 and the three states, there are 15 observations in the panel data model. However, the model can be tweaked in a more focused way to understand how the SBM and NBA affected people and consequently – the impact of the program on disease. For the
  • 5. 5 |C h u NBA, the program paid households subsidies to build the toilets in their own homes. The government offered 4600 rupees to any homeowner, especially poor home owners, to construct their own individual toilets.6 As a result, there exists an interaction factor between individual toilet constructions because building a toilet has the cash incentive which may make the homeowner more likely to begin construction of the toilet, consequently, impacting their disease instances. Due to the multicollinearity of GDP per capita and the interaction term of the continuous variable of GDP per capita and toilets constructed, the variable for GDP per capita is dropped out and exchanged for the interaction term. Model 2 - New regression with interaction term Data For the 15 panel observations, most of the data falls into a relatively reasonable range when applying the natural logarithmic transformation; unfortunately, when trying to expand the observations by increasing the time scale from annual to quarterly or monthly is not applicable as many of the observations fluctuate in random patterns as the construction numbers are delayed in their reporting. Subsequently, the same problem emerges when attempting to use district level data as the other data such as cases of acute diarrhea or literacy rate are not accurately reported for each year. Therefore, using the annual data is the only way to insure validity for the measurements within the data. Unfortunately, when attempting to run the extended fixed effects regression model, the increased number of independent variables minimized the observations in favor of degrees of freedom used in the F-test which is a measure to illustrate the fit of a regression model on the data. 6 “Restructuring of the Nirmal Bharat Abhiyan into Swachh Bharat Mission,” Press Information Bureau, Government of India, Sep 24 2014.
  • 6. 6 |C h u Looking at Figure 5, while the absolute totals of the variables do suffer from occasionally large variances and, consequently, high standard deviations, they still form only a small fraction of the mean in most cases meaning the points do not differ too much as a proportion of the mean. That point can be illustrated in examining the kurtosis of the tails which indicate the amount of outliers within the observations; most of these points do not exceed the “baseline” value of 3 implying the tails are not oversaturated with observations which would hamper the OLS assumptions of the model. For the total values, especially for the construction numbers, one must expect a rather large fluctuation between the points given the differences between states which will be covered more in the model results. Looking at Figure 6, one can see that the dependent variable (“lntotalcases”) does not have a very large standard deviation which is indicative of the logarithmic transformation as the natural logarithm requires rather large values to fluctuate in value; however, it is also indicative that the percentages of those contracting these symptoms do not vary greatly from year to year. The same can be said for the independent variables (“lntoilets”, “lnBPLtoilets”, “lnGDPPC”, “lev1rurallit, lnsharespent”). In addition, the kurtosis and skewness of each variable except “lnsharespent” reflects a relatively normal distribution of the points indicates there are no strong outliers which show strength in the data. Case Studies The main analytic lens applied for overall success for these types of programs in a similar situation will be the implementation of the toilet construction initiative in Indonesia; the crucial paper is a working paper written by the World Bank entitled Impact Evaluation of a Large-Scale Rural Sanitation Project in Indonesia written by Lisa Cameron, Manisha Shah, and Susan Olivia (hereafter referred to as Cameron et al.) The writers studied eight rural districts in 160
  • 7. 7 |C h u communities in Indonesia attempting to study the Total Sanitation and Sanitation Market (TSSM) project.7 While the toilet construction in the treated communities increased by three percentage points and diarrhea prevalence was only 3.3 percent in the treated areas compared to 4.6 in those that were not, it did not impact sanitation habits such as washing hands, open defecation, or using soap which also have a large impact on diarrhea conception.8 When measuring certain baseline demographic descriptive statistics such as defecting in an improved sanitation facility, in the open, washing hands after going to the toilet, and having soap, none of the respective difference-in-difference p-values are statistically significant except for the construction of toilet variable, for obvious reasons.9 Unfortunately, that means that the program’s efficacy was not explicitly tied to toilet construction but the decreases in diarrhea in the control area may be correlated to other factors outside of the scope of the TSSM. For the SBM, this factor is important in attempting to discern the impact of toilets and knowledge of good sanitation practices by itself. The supplementary case study is the piece written by Greenstone and Hanna in which they discuss the efficacy of the Indian government attempting to rein in air and water pollution: the Supreme Court Action Plans and Mandated Catalytic Converters, as well as the National River Conservation Plan.10 Through a Quandt likelihood ratio test, the researchers found that there was almost no impact from the adoption of the National River Conservation Plan (NRCP) but a definitive structural break from the air pollution policies from which very changes emerged as an engaged public galvanized the additional push needed to execute the policy despite less 7 “Impact Evaluation of a Large-Scale Rural Sanitation Project in Indonesia,” Cameron et.al, The World Bank, Water and Sanitation Program, Feb 2013. 8 Ibid. 9 Ibid. 10 “Environmental Regulations, Air and Water Pollution, and Infant Mortality in India,” Greenstone et al, Massachusetts Institute of Technology, Feb 2011.
  • 8. 8 |C h u than efficient government structures.11 The researchers concluded that this break emerged from a factor of utility juxtaposed with more energized and efficient institutions (the Supreme Court) that could effectively legislate and dole out proper regulations and create enforcement mechanisms.12 In addition, the study showed that the efficacy of the program was tied to the high avoidance costs for air pollution as people can only breathe air in a certain medium and cannot buy imported air while water is more easily cleaned through boiling it.13 The main conclusion from these points to the eventual success or failure or the SBM is that the government policies can succeed when the populace stands behind the program and encourages their own part of the policy. Another conclusion is that the program functions best when the avoidance costs are high and people feel a definitive change in their avoidance of the pollutants, especially in a comparison between water and air pollutants. Results While much of the program is rather new, the results gathered were a mixed affair in terms of expectations. Before the experimentation with the model, many of the stories regarding the efficacy of the SBM stated that many of those who had toilets constructed in their homes or local townships felt too constrained to use the new toilets provided to them and still chose to defecate in the open. Due to these untraceable biases, many of the independent variables lack the statistical significance needed to assert their own validity as variables. While it is possible that the lack of significance for some of these variables is due to the overall lack of observations, looking at the graphs in Figure 2 clearly shows trends that would support the present hypothesis that increased facilities and increased pressure to use them will decrease the instances of diarrhea 11 “Environmental Regulations, Air and Water Pollution, and Infant Mortality in India,” Greenstone et al, Massachusetts Institute of Technology, Feb 2011. 12 Ibid. 13 Ibid.
  • 9. 9 |C h u in these poor states. For example, the regression results in Figure 3 using the robust standard error assumption still show the same relative t-statistic which implies that these observations are relatively consistent even if the statistical significance is rather small compared to the desired level. However, with that being said, there is clearly some form of bias that is accounting for the fluctuations in the total instances which is clearly encapsulated in the variables marking the state id in both Figure 1 and Figure 3 as state id 2 (Madhya Pradesh) and state id 3 (Rajasthan) are statistically significant to a reasonable degree. In addition, the variable for GDP per capita measured a positive variable, meaning that increases in overall GDP per capita may be detrimental to health; however, one must consider the impact of the subsidies on the toilet and see if that extra incentive at all impacts the overall levels of diarrhea. While looking at the other model comparing the interaction factor of GDP per capita and total below poverty line toilets constructed in a logarithmic transformation in Figure 4, there is a clear significance level of both the construction of toilets below the poverty line and the subsidy put in place for that construction. Unfortunately, the signs for both variables are negative and positive respectively meaning that the impact of the subsidy for each toilet built is negatively increasing or time. Thus, the model implies that the impact of toilet building with the subsidy, which is already built into the legislation, the toilets themselves are only effective to a certain point in a case of diminishing margins of returns. While many of the independent variables in this regression lack overall statistical significance, it is important to include them as they serve to eliminate any omitted variable bias. Whether or not the overall independent variables are statistically significant, the individual state levels must also be considered in order to gain the full picture.
  • 10. 10 |C h u In analyzing both Model 1 and Model 2, there is a clear statistical significance in both Madhya Pradesh (state 2) and Rajasthan (state 3) as opposed to Uttar Pradesh (constant). Additionally, the signs on both of the states are negative which indicates that there are underlying factors within each state that are not necessarily outlined in the model which help lower the percentages of diarrhea occurrence. Therefore, there is some unknown factor or factors in the model which shows that the states Madhya Pradesh and Rajasthan are healthier than Uttar Pradesh beyond the variables stated in the two different regressions. Conclusions and Policy The program push for a “Clean India” is certainly not without its warts or inefficiencies; however, the impact of toilets and toilet usage is still yet to be determined under the SBM and NBA. Unfortunately, it is clear that toilets are not the only factor in overall health due to the fact that they are not necessarily the biggest variable factor in the models; this news is of no real surprise given the lessons provided in Indonesia with their TSSM program. While looking at the stories surrounding the program itself, the reception has not been incredibly outstanding. In an interview with Kartikay Mehrotra of Bloomberg Business, a woman named Sunita in Uttar Pradesh stated that “only Dalits, the lowest Hindu caste, should be exposed to excrement in a closed space.”14 That mentality is the main problem in the overall longevity of the program; without people being willing to use the toilets, they are effectively useless in preventing instances of diseases. As was the case in Indonesia, one can lead the program and build all the sanitary conditions for toilets as necessary but they are not useful when left idle. As stated by Archana Patkar, a program manager at the Water Supply & Sanitation Collaborative Council, “‘ the problem has gotten worse with the government thinking this is a supply driven problem. The 14 “India’s Toilet Race Failing as Villages Don’t Use Them,” Mehrotra, Kartikay, Bloomberg Business, Aug 4, 2014. Web.
  • 11. 11 |C h u problem is that germs are invisible, and so understanding the threat of open defecation is far removed from reality.”15 Therefore, the next step is doing internal research on the mentality surrounding toilet usage to enlighten people to the changes that clean defecation practices can have in their lives. Without properly addressing the mentality around the usage of toilets as something only meant for lower strata of society, there is only so much that can be accomplished by building toilets without any education. For example, when the World Bank released their study on Indonesia, surveys on good sanitary behavior were released to both the treatment and control group on their understanding of these topics; these results showed that the changes in construction of sanitary areas did not create a significant impact on beliefs around proper sanitation.16 By educating people below the poverty line, as the caste is not necessarily tied to economic status, there can be a real impact of using those toilets on diarrhea death and diseases; however, as the old saying goes, you can lead a horse to a toilet but you cannot make it use the toilet. 15 “India’s Toilet Race Failing as Villages Don’t Use Them,” Mehrotra, Kartikay, Bloomberg Business, Aug 4, 2014. Web. 16 “Impact Evaluation of a Large-Scale Rural Sanitation Project in Indonesia,” Cameron et.al, The World Bank, Water and Sanitation Program, Feb 2013.
  • 12. 12 |C h u Figures: Figure 1 – fixed effects regression using below poverty line household toilet construction measurements. Figure 2 - Graphs showing the changes in total cases of diarrhea and other indicator variables measured by state id (1- UP, 2 - Madhya Pradesh, 3- Rajasthan) _cons 3.836265 6.993411 0.55 0.598 -12.29057 19.9631 _Istate_id_3 -1.325917 .5417518 -2.45 0.040 -2.575199 -.0766347 _Istate_id_2 -.4617604 .4158162 -1.11 0.299 -1.420634 .4971135 lnGDPPC 1.230642 .6481366 1.90 0.094 -.2639639 2.725248 lev1rurallit .0206392 .0965196 0.21 0.836 -.2019354 .2432137 lntotalsancov -.1230762 .090466 -1.36 0.211 -.3316911 .0855387 lnBPLtoilets -.19465 .1377828 -1.41 0.195 -.5123778 .1230778 lntotalcases Coef. Std. Err. t P>|t| [95% Conf. Interval] Robust Root MSE = .27938 R-squared = 0.7461 Prob > F = 0.0002 F(6, 8) = 21.29 Linear regression Number of obs = 15 i.state_id _Istate_id_1-3 (naturally coded; _Istate_id_1 omitted) . xi: regress lntotalcases lnBPLtoilets lntotalsancov lev1rurallit lnGDPPC i.state_id, r
  • 13. 13 |C h u Figure 3 - Same regression as Figure 1 but with robust heteroskedastic errors Figure 4- Same regression as Figures 1 and 3 but including a subsidy interaction term (which drops out GDP variable) _cons 1.40949 6.554861 0.22 0.835 -13.70605 16.52503 _Istate_id_3 -1.439068 .49245 -2.92 0.019 -2.57466 -.3034764 _Istate_id_2 -.5625891 .3866586 -1.46 0.184 -1.454226 .3290473 lnsharespent -.1044713 .0528974 -1.97 0.084 -.226453 .0175104 lnGDPPC 1.310905 .5904148 2.22 0.057 -.0505935 2.672404 lev1rurallit .0103106 .098293 0.10 0.919 -.2163535 .2369747 lnBPLtoilets -.1013052 .1005108 -1.01 0.343 -.3330836 .1304732 lntotalcases Coef. Std. Err. t P>|t| [95% Conf. Interval] Robust Root MSE = .26847 R-squared = 0.7656 Prob > F = 0.0004 F(6, 8) = 16.82 Linear regression Number of obs = 15 i.state_id _Istate_id_1-3 (naturally coded; _Istate_id_1 omitted) . xi: regress lntotalcases lnBPLtoilets lev1rurallit lnGDPPC lnsharespent i.state_id, r _cons 3.836247 6.993453 0.55 0.598 -12.29068 19.96318 _Istate_id_3 -1.325916 .5417537 -2.45 0.040 -2.575203 -.0766302 _Istate_id_2 -.461761 .4158177 -1.11 0.299 -1.420638 .4971164 lnGDPxBPLtoilets 1.230642 .6481394 1.90 0.094 -.26397 2.725254 lev1rurallit .0206393 .09652 0.21 0.836 -.2019362 .2432148 lntotalsancov -.1230756 .0904663 -1.36 0.211 -.3316912 .08554 lnBPLtoilets -1.425292 .7257946 -1.96 0.085 -3.098977 .2483936 lntotalcases Coef. Std. Err. t P>|t| [95% Conf. Interval] Robust Root MSE = .27938 R-squared = 0.7461 Prob > F = 0.0002 F(6, 8) = 21.29 Linear regression Number of obs = 15 i.state_id _Istate_id_1-3 (naturally coded; _Istate_id_1 omitted) > ate_id, r . xi: regress lntotalcases lnBPLtoilets lntotalsancov lev1rurallit lnGDPxBPLtoilets i.st
  • 14. 14 |C h u 99% 660.3888 660.3888 Kurtosis 5.810532 95% 660.3888 373.6622 Skewness 1.712458 90% 373.6622 257.7996 Variance 26962.84 75% 256.8474 256.8474 Largest Std. Dev. 164.2036 50% 144.026 Mean 183.3494 25% 56.7074 56.7074 Sum of Wgt. 15 10% 43.5264 54.2441 Obs 15 5% 3.75 43.5264 1% 3.75 3.75 Percentiles Smallest sharespent 99% 72391.2 72391.2 Kurtosis 2.136755 95% 72391.2 65658.42 Skewness .5419822 90% 65658.42 61480.77 Variance 1.94e+08 75% 58636.92 58636.92 Largest Std. Dev. 13943.92 50% 42496.59 Mean 45256.68 25% 33366.44 33366.44 Sum of Wgt. 15 10% 30038.24 32087.83 Obs 15 5% 26539.93 30038.24 1% 26539.93 26539.93 Percentiles Smallest GDPPC 99% 2915407 2915407 Kurtosis 6.888059 95% 2915407 1613384 Skewness 2.03895 90% 1613384 1166016 Variance 4.74e+11 75% 900769 900769 Largest Std. Dev. 688177.2 50% 648792 Mean 818640.1 25% 515427 515427 Sum of Wgt. 15 10% 252800 266197 Obs 15 5% 134873 252800 1% 134873 134873 Percentiles Smallest totaltoilets 99% 1042578 1042578 Kurtosis 4.097271 95% 1042578 711103 Skewness 1.377213 90% 711103 621743 Variance 7.53e+10 75% 472521 472521 Largest Std. Dev. 274358.4 50% 213312 Mean 328844.5 25% 158175 158175 Sum of Wgt. 15 10% 81700 102905 Obs 15 5% 45359 81700 1% 45359 45359 Percentiles Smallest totalBPLtoilets 99% 826246 826246 Kurtosis 2.096739 95% 826246 768021 Skewness -.2490236 90% 768021 745457 Variance 3.66e+10 75% 740328 740328 Largest Std. Dev. 191327.5 50% 535012 Mean 539293.5 25% 431893 431893 Sum of Wgt. 15 10% 227571 290705 Obs 15 5% 223106 227571 1% 223106 223106 Percentiles Smallest totalcases Figure 5- Summary Statistics of all dependent and independent variables in absolute form
  • 15. 15 |C h u 99% 6.492829 6.492829 Kurtosis 5.626488 95% 6.492829 5.923352 Skewness -1.469123 90% 5.923352 5.552183 Variance 1.445991 75% 5.548482 5.548482 Largest Std. Dev. 1.202494 50% 4.969994 Mean 4.76409 25% 4.037905 4.037905 Sum of Wgt. 15 10% 3.773368 3.993494 Obs 15 5% 1.321756 3.773368 1% 1.321756 1.321756 Percentiles Smallest lnsharespent 99% 16.2 16.2 Kurtosis 2.161826 95% 16.2 15.9 Skewness .2154488 90% 15.9 15.1 Variance 2.265524 75% 15.1 15.1 Largest Std. Dev. 1.505166 50% 13.3 Mean 13.58667 25% 12.7 12.7 Sum of Wgt. 15 10% 11.4 12.4 Obs 15 5% 11.2 11.4 1% 11.2 11.2 Percentiles Smallest lev1rurallit 99% 11.18984 11.18984 Kurtosis 1.938605 95% 11.18984 11.09222 Skewness .1607812 90% 11.09222 11.02648 Variance .0917857 75% 10.97912 10.97912 Largest Std. Dev. .3029615 50% 10.65718 Mean 10.67694 25% 10.41531 10.41531 Sum of Wgt. 15 10% 10.31023 10.37623 Obs 15 5% 10.18641 10.31023 1% 10.18641 10.18641 Percentiles Smallest lnGDPPC 99% 13.85721 13.85721 Kurtosis 2.60935 95% 13.85721 13.47457 Skewness -.1402715 90% 13.47457 13.34028 Variance .700332 75% 13.06584 13.06584 Largest Std. Dev. .8368584 50% 12.27051 Mean 12.39531 25% 11.97146 11.97146 Sum of Wgt. 15 10% 11.31081 11.54156 Obs 15 5% 10.72236 11.31081 1% 10.72236 10.72236 Percentiles Smallest lnBPLtoilets 99% 13.62465 13.62465 Kurtosis 2.629621 95% 13.62465 13.55157 Skewness -.8406101 90% 13.55157 13.52175 Variance .1756777 75% 13.51485 13.51485 Largest Std. Dev. .4191393 50% 13.19004 Mean 13.1254 25% 12.97593 12.97593 Sum of Wgt. 15 10% 12.33522 12.58006 Obs 15 5% 12.3154 12.33522 1% 12.3154 12.3154 Percentiles Smallest lntotalcases Figure 6- Summary Statistics of all dependent and independent variables in natural logarithm form
  • 16. 16 |C h u References: 1. “ASER Yearly Reports”, Aser Centre, Pratham Education Foundation, http://asercentre.org, Web. 2. “CHBI National Health Profiles”, Central Bureau of Health Intelligence, http://cbhidghs.nic.in , Web. 3. “Environmental Regulations, Air and Water Pollution, and Infant Mortality in India,” Greenstone et al, Massachusetts Institute of Technology, Feb 2011. 4. “Impact Evaluation of a Large-Scale Rural Sanitation Project in Indonesia,” Cameron et.al, The World Bank, Water and Sanitation Program, Feb 2013. 5. “India’s Toilet Race Failing as Villages Don’t Use Them,” Mehrotra, Kartikay, Bloomberg Business, Aug 4, 2014. Web. 6. “India’s Toilet Race Failing as Villages Don’t Use Them,” Mehrotra, Kartikay, Bloomberg Business, Aug 4, 2014. Web. 7. “National Rural Drinking Water Programme”, Government of India, http://indiawater.gov.in , Web. 8. “PM Modi fulfills promise of 80 lakh toilets but not many takers in rural India,” Sharma, Neetu: India Today, Aug 14, 2015. Web. 9. “Restructuring of the Nirmal Bharat Abhiyan into Swachh Bharat Mission,” Press Information Bureau, Government of India, Sep 24 2014. 10. “Swachh Bharat Mission”, Government of India, http://tsc.gov.in , Web.