Paper Submission
The paper must pose a question and seek an answer. All papers must follow the same
format given below, problem statement, motivation, analytical framework, data sources
and findings.
The paper proposal must contain 1 and 2.
Word Limit: 2500 words excluding references, graphs and tables. Embed the graphs,
tables in the text (i.e., don’t push them to the end).
The paper must be accompanied by a slide deck – max 10 slides. In grading, I will first
look at the slides.
Format for file name: Your last names paper and Your last names slide.
I explain them below with a completely fictious example.
The examples are merely illustrative. There are uncountably many ways in which to write
the paper. Be engaging, be interesting and above all, be thoughtful.
Option #1: Data Analysis
1. Problem statement (10%): A short 3-4 line problem statement which states the question
that the paper will answer.
Example: I will examine the relation between a child’s height-for-age and the mother’s
education
2. Motivation (25%): Here the paper must justify the question. Why is it a problem? Why
do we care about the answer? In other words, what is interesting about the question?
Example: India has the world’s largest proportion of stunted children (cite references or
show data). Many explanations have been proposed for it – including poor diets, disease
environment, and poor health of mothers leading to low birth weight (cite references,…).
Here I wish to explore a different mechanism – could it be that parents are unaware of the
good diets and hygiene that lead to better health outcomes for their children?
Prior work has investigated this question in some countries in Africa and Central America
(references). While Rabada [Link] find a significant impact of information campaigns on using
chlorine tablets to purify water, Ramirez and Eisen find weak impacts of social media
campaigns on breast feeding.
In this paper, we revisit this issue in the context of India.
3. Analytical Framework (25%): Include theory and empirical methods here. Write the
regression specification. If it is causal, state why so or under what assumptions. Mention
limitations, if any, regarding causality
Example:
Parents choose diets and hygiene practices based on their perceptions of benefits and costs.
Information alters certain components of benefits and costs. For instance, information about
using chlorine tablets increases the benefits of that action while increasing the costs of using
untreated water (by attributing diarrhea episodes to this practice).
In this paper, because of data limitations, we do not evaluate information campaigns. Instead,
we use the mother’s education as an indicator of the parent household’s capability to process
and understand information about diets and health practices. We look at its relation to a
child’s height-for-age – a measure of the child’s health.
We run a regression of the following form for the period 2546-2555. To minimize spurious
correlation, the control variables are the following – income, household size, # of siblings,
blah-blah. The survey records levels of education (primary, secondary, high school etc)
rather than years of schooling. Therefore, we use dummy variables for these levels.
Since mother’s education is not randomly assigned, there is still a possibility that our results
are not causal because of household characteristics that we are unable to observe.
4. Data Sources (10%); Describe the data source, the variables you use and put up a table of
descriptive stats.
E.g., Our data comes from 8th generation NEO chips implanted in every household member
under the DYST program launched by the government in 2550. The data relates to a random
sample of 3 million households in the state of X. We have information on household
demographics, income, food intake, and heights of children (under 15 years of age).
The descriptive statistics of the variables used in the analysis are displayed in Table ?. As
can be seen the majority of mothers in the sample have education levels lower than high
school. About 40% of children are shorter than the relevant WHO benchmark.
5. Findings (30%): Interpret your findings, what they mean and mention policy implications,
if any.
E.g.,Table 28 displays the regression results. In the first column, we have the results without
control variables. These are added in the second column. The third column adds planet fixed
effects.
It can be seen that results on mother’s education are stable between specifications. The
standard errors are little higher in column 3.
On the whole, mother’s education does matter. Relative to those who have only a primary
education, height-for-age is about 0.3 standard deviations higher for those with secondary
education and 0.37 standard deviations higher for those with high school education (column
3). The p-values for these coefficients are less than 0.5. We are unable to reject the null that
the coefficients for high school graduates and college graduates are equal. In other words,
the mother’s education ceases to have an impact beyond high school education…….
We know that drop out rates are the highest for girls after primary education. That means
favourable impacts on child health in future generations are lost. There are many good
reasons why the government should strongly move to maximize school completion for girls.
The effects estimated in this paper provide yet another argument.
Option #2: Survey of Literature
A second option is to survey a minimum of three papers. The papers can be research or
policy papers. It can also include country case studies.
This should follow the following format.
1. Problem Statement: This must pose a question exactly like in Option #1.
e.g., Is child labour a serious problem in our state?
2. Motivation: Like in Option #1. It would be good if at least came up with some numbers
and descriptive stats to justify the question. Of course, you can also talk about some prior
findings that lead to this question.
E.g., from previous studies, we know that child labour is used in some specific sectors – like
dhabhas, fireworks manufacture, carpet manufacture and as domestic servants. This happens
even when the economy is growing well. These children are being deprived of their
education, health, and of course their childhood – unfreedoms as Amartya Sen would call it.
It is important that we understand the situation in our state in order to understand the policies
that can reduce and eliminate child labour entirely.
3. Analytical framework: Speak about how one would analyze it and possible difficulties in
doing so.
E.g., the magnitude of the problem depends on (a) the numbers of such children (b) the age at
which they are employed (the lower it is, the worse is the deprivation) (c) the conditions of
work – what hazards are they exposed to (d) do they live with their parents? If not, do we
know if they are subject to abuse? (e) do they work full-time or are they combining school
work with a job?
Then you can say we are going to look at these zillion papers that explore some of these
dimensions.
Or you could say more – that you look at studies that study this in different ways – one is a
case study of a neighbourhood, another is a cross-section at a point in time, yet another study
is longitudinal as it follows a bunch of children over time as they grow into adults etc etc
Or that they relate to different sectors – or different areas.
Point is you must justify the selection – why these papers? What is interesting/important
about them?
4. Description of the papers – chose some dimensions along which you wish to describe –
the time period, the households covered, the methods followed – anything that will illuminate
the problem at hand. You could have a table that describes these things and keep the section
short.
E.g., A is an analysis of Census data on the prevalence of child labour, B is a regression
analysis of the effect of family income on the incidence of child labour, C quantifies in
monetary terms the loss of human capital because of child labour, D is a case study of the
conditions of work of children employed as domestic servants….
5. Findings, discussion, policy implications
Don’t catalogue the findings – X says this, Y says that, etc etc. That is not insightful.
You must have a narrative where each paper segues into another.
E.g., begin with study A about a macro picture of child labour – what age group, how many,
what proportions, how many years of such work etc. Move to study B about the determinants
of incidence – move to study D about the working conditions and the deprivations faced by
children and finally to study C which quantifies the monetary loss.
Remarks
I am interested in how thoughtfully you bring the materials (data, analysis, other literature) to
answer the question at hand.
If you can relate it to what we study in class, that would get you credit. if you have learnt x, y
and z in this class, how does the new material deepen, modify, reinforce or contradict that
understanding? The answer here would be your paper’s contribution.