Does GDP per Capita Make Countries Happy? There’s an old saying that “money cannot buy happiness”, but how true is this statement in reality? In this project I want you to perform a series of cross- country regressions that regress national-level subjective well-being scores on GDP per capita. With the data provided use Stata to perform an analysis that addresses the above question and write up yout results in the form of a report.

QUESTION

Does GDP per Capita Make Countries Happy?
There’s an old saying that “money cannot buy happiness”, but how true is

this statement in reality? In this project I want you to perform a series of cross-
country regressions that regress national-level subjective well-being scores on

GDP per capita. With the data provided use Stata to perform an analysis that
addresses the above question and write up yout results in the form of a report.
This is a reasonably open question and thus offers you a platform to showcase
both your data management and econometric/Stata software skills alongside
your ability to motivate, interpret, and understand econometric procedures.

Your answer to this question will be graded according to Queen’s Concep-
tual Equivalence scale. Please see the following url for further details:

https://www.qub.ac.uk/directorates/media/Media,837251,smxx.pdf
Suggested Report Structure
Introduction
A brief introduction to the problem and provide a short overview of how
you will answer the question. What factors do you think will be important
and why? What kind of issues do you think an econometrician faces when
analysing this question?
Don’t be afraid to read further on this topic and cite relevant literature, where
appropriate.
A few paragraphs will suffice. Less than one page.

1

Data

A description of the data used in the analysis. There is nothing in the assign-
ment that requires you to work with all of the data provided. If I was per-
forming this analysis I would only focus on several variables. You will need

to decide yourself on how many variables you want to include in the analysis.
A summary statistics of the variables and perhaps one or two plots (such as
a scatter-plot showing the relationship between GDP per capita and subjective

happiness) showing some basic information and relationships would be ap-
propriate in this section. A few pages here, say between 2 to 4, is probably

OK.
Analysis and Results
Use this section to present your regression results. I would advise that you run
a variety of models. This is not an assignment where you are required to get

one specific answer like “GDP per capita has a x% effect on subjective well-
being…”. Instead, the effect will obviously depend on the context, so a better

analysis will be able to present a range estimates and motivate and interpret
them.
For example, you might find that the estimated slope associated with GDP
per capita is smaller in magnitude when you include an additional explanatory
variable measuring Press Freedom. This would indicate that at least part of
the bivariate income-happiness relationship is driven by freedom in the press.
People are happy that they live in countries where the media is not heavily
censored and also these countries have higher levels of GDP per capita….
You only need to mention the approach or procedure you are using and
don’t need to tell me computational details (assume I know this). For example,
don’t write “to estimate the coefficients we perform the OLS computation using
matrix algebra βˆ = (X
0X)
−1X
0Y…” or don’t tell me how to do a t-test when I

can see the outcome of the test in the regression table.
It is difficult for me to recommend a page or word limit here. In the past
students have written a lot and used a lot of pages but their answers have been
poor. On the other hand I have corrected some excellent short answers. Part of
what you are being graded on is your ability to succinctly interpret regression
model results. One recommendation I would make is to not clog the report
up with regression tables that only show coefficients from single regression
models. Instead use the append option in outreg2 to generate regression tables
that provide information for multiple regression models.1
1Obviously you will need to exercise some degree of restraint here. For example, don’t try

2

Conclusion
Re-iterate your findings and touch on some shortcomings in your analysis and
offer potential avenues for future research. Like the intro, a few paragraphs
will do.
Do-File Appendix
A working do file must be appended to your work. If you do not append a do
file with your project, your work will be deemed incomplete, and you will be
asked to resubmit with the relevant penalty for late submission being applied.
Working do file means that the do file should contain all of the code used to
generate your results. This also includes setting the working directory, loading
the data, and saving graphs and tables etc. If any code is omitted, the project
will be treated as incomplete.
In practical terms, this means that you need to copy any paste the do file
commands into your Word file containing you project write up at the end of
this project.
The do file appendix does not count towards either the word count or page
limit.

The Assignment Data

The data you will use has been adapted from the Quality of Government Insti-
tute:

https://qog.pol.gu.se/data/data-portal.
Each observation in these data represents a different country. When you open

these data you will see the variable’s names and labels. Most of the vari-
able’s contents are self-explanatory. Detailed instructions of the variables can

be found in the dataset codebook available at the following link:
https://www.qogdata.pol.gu.se/dataarchive/qog_std_jan19.pdf
Generating or Reforming Variables
You should also think in terms of how variables should be expressed to aid
your analysis. Should you variables be transformed (i.e. log or square root) to
aid interpretation? Are quadratic forms (squared) of the variables appropriate?
and squeeze in a table of regression coefficients from 8 models. In that instance 2 tables with
4 models would be a better way to display your results.
3

What about interaction variables? The lecture notes and slides should provide
you with sufficient guidance on this. There is also nothing stopping you from
creating new variables. For example, you may wish to generate a dummy
variable for all countries who are in a Europe and so on.
How Do I Make my Report Great?
You are required to run regression models of the form:
Subjective Well-Beingi = [Intercept] + [Explanatory Variables] + [Error]
where “Subjective Well-Being” and “Explanatory Variables” refer to variables
provided in these data or a transformed measure that might make it easier for
you to interpret the explanatory coefficients (logs perhaps?). At the most basic
level you should be able to select a sample, run some regressions, display the
results in a table, and tell me what they mean.
A more complete answer should have you running multiple specifications
with different explanatory variables and discussing how and why the results

change. Remember there is no one true regression specification you are to un-
cover. A good idea is to use the “append” function from “outreg2” to display

different specifications side-by-side and thus save space. You don’t need to

shove all of your regressions into one table either—be mindful of the presenta-
tion aspect.

You should be comfortable working with these data and generating and
transforming variables where required to thus improve the exposition of your
report.

I would also expect a “good” answer to contain some form of critical el-
ement that shows you have considered, and thus flagged in your report, the

potential flaws in the linear regression approach and have either attempted to
correct for them or have a suggestion on how to correct for them in future
research.

More Hints and Suggestions
Make your output as tidy and professional as possible. I suggest you save
your graphs as .png images using Stata’s graph export command.
Use Stata’s outreg2 function to make tables of your summary statistics
and regression results. Use the append feature if you want to save space and
display multiple regression results.

4

Feel free to edit the regression tables as you wish, especially if you think it
will make your output easier to interpret. However, the following information
should always be included in a regression table: all coefficient estimates, their
standard errors, the number of observations, and the R
2
statistic.

Label all tables and figures.

5

Instructions and Regulations
Groups
I am assuming that you are working in the same group as the first project.
Please inform me of any group changes as soon as possible. The maximum
group size is 4. DO NOT ASK ME IF YOUR GROUP CAN CONSIST OF
MORE THAN FOUR PEOPLE.
Deadline
Each project will be completed and submitted to Canvas before 5:00pm on the
27th of March 2020 at the latest. There are two project assignment groups on
Canvas and one person from your group will have to submit to both.
Formatting Instructions
The project should be typed in 12 point font size, double spaced, on A4 paper,
and should not exceed 1,500 words length (excluding graphs, tables, equations,
and bibliographical references). This word count is an upper limit. The whole
project, everything included but excluding the cover page, must not exceed 15
pages (5% deducted for each additional page). Again this is an upper limit.

Every page should have the page number printed in the middle at the bot-
tom.

The project should contain a cover page (not numbered) with the following
information:
• Title
• Full name(s) and degree(s)
• Email address(es)
• Student ID number(s)
Each project should have a References section if works are cited. Each
answer must have the same number as the corresponding question and must
be presented in the same order as in the project. Make sure each graph’s axes
are clearly labelled. Do not include graphs or tables on which you do not
comment in the report. Marks will be deducted for a messy presentation of
results. See “Hints and Tips” for further details.

6

References
If you need to refer to any results established in another person’s work, clearly
indicate the reference to that work by writing down the family name(s) of the
author(s) and the year of publication—for example, Fernihough (2015). Make
sure that you indicate the full reference to the work in the References part of
your project. Use the Harvard referencing scheme.
Late Submission Penalties and Exemptions
The Study Regulations For Undergraduate Programmes is available at the url listed
in the below.
These regulations discuss late penalties for submitting assessed work after
the deadline.
https://www.qub.ac.uk/directorates/AcademicStudentAffairs/AcademicAffairs/
GeneralRegulations/StudyRegulations/StudyRegulationsforUndergraduateProgrammes/

These regulations also provide further information on the exceptional cir-
cumstance procedures. Further information about this process is contained at

the below urls:
https://www.qub.ac.uk/directorates/AcademicStudentAffairs/AcademicAffairs/
StudentGuidance/ExceptionalCircumstances-AStudentGuide/
Plagiarism, Duplication, Collusion and Fabrication
Carrying out, and writing up the project must be exclusively the work of each
group. Plagiarism, duplication, collusion (between groups rather than within
groups) and fabrication is of concern to the university and your work should
conform to the requirements of academic integrity. All suspect assignments
will be reported without exception. You have been warned.
Hints and Tips
Part of this project will be assessed on the project formatting and how neat
and tidy your report is. In addition to the formatting and other instructions
outlined in the above I suggest you obey the following rules when writing your
project up.
1. Justify the text in MS Word (not align right), or whatever word processor
you use.
2. Label all tables and figures appropriately. Table 1, Table 2, Figure 1, and
so on.

7

3. Tables are tables, not figures, so label them correctly.
4. Double space text.
5. Don’t use any weird font type. Ariel, Times New Roman, and Calibri are
standard.
6. Don’t tell me what you did. For example, “I turned on my PC and loaded
Stata. After having a cup of tea I opened the data file and began to work,
and so on”.
7. Never copy and paste a screenshot. Edit your tables as instructed and
make them tidy.
8. Make sure your tables are consistent, i.e. things are on the right lines and
they are easy to read and interpret.
9. Put your labels at the top of each table and figure. For example, “Table 1:
Summary Statistics” should be at the top of the table.
10. Don’t pad your analysis out to meet the max word/page counts. These
counts are set as upper limits.
11. Pay attention to the number of decimal places you use when quoting a
number. More is not better.

8

Table 1: Some Stata Commands and Operators and Their Use
Code Use

clear clears the Stata workspace, any previ-
ously loaded data is removed

cd “C:\econometrics” Sets the working directory to a folder
called econometrics in the C: drive
use “data.dta” Opens a Stata dataset called data
generate y=1/x Create a new variable called y which is the

inverse of the variable x

replace y=1/x Replace an existing variable called y with

the inverse of the variable x

drop x Remove the variable x from dataset loaded

into Stata

drop if x>0 Remove all observations from the dataset
for which the variable x is greater than zero
summarize x y Get the summary statistics for the variables

x and y

outreg2 using
summarystats.doc,
replace sum(log) keep(x
y)

Word output for summary statistics

graph box y, over(x) Get a box plot of the variable y stratified

by the variable x

regress y x1 x2 Regress the variable y on the variables x1
and x2, i.e. run the regression yi = β0 +
β1x1i + β2x2i + ei

regress y x1 x2 if z==0 Regress the variable y on the variables x1
and x2 but only for observations where z
is equal to zero

outreg2 using reg1.doc,
replace ctitle(Model 1)

Create a Word document in your working
directory called reg1.doc that contains the
results of the previous regression

outreg2 using reg1.doc,
append ctitle(Model 2)

Modifies an existing Word document in
your working directory called reg1.doc
and creates a new column of results beside
the previous ones

predict yhat, xb Create a new variable that contains the

predicted values (yˆi

) from a previously run

OLS regression

ANSWER

The Relationship Between GDP per Capita and Subjective Well-Being: An Econometric Analysis

Introduction

The question of whether GDP per capita influences subjective well-being is a complex and important one. This report aims to investigate this relationship through a series of cross-country regressions. The analysis will provide insights into the factors that may be important for happiness and shed light on the challenges faced by econometricians when examining this question. Relevant literature will be considered to provide a comprehensive overview.

Data

The analysis will utilize data from the Quality of Government Institute. The dataset includes variables related to subjective well-being, GDP per capita, and other potential explanatory variables (OECD Publishing, 2013). A selection of variables will be chosen for analysis, considering their relevance and availability. Summary statistics will be presented, and scatter plots will be generated to explore the basic relationships between GDP per capita and subjective happiness.

Analysis and Results

Multiple regression models will be estimated to examine the relationship between GDP per capita and subjective well-being. Various specifications will be explored by including additional explanatory variables that could potentially influence well-being (Proto & Rustichini, 2013). The regression results will be presented and interpreted, highlighting the estimated coefficients, standard errors, the number of observations, and the R-squared statistic.

The analysis will also consider potential interactions between variables and explore transformations or quadratic forms of the variables to aid interpretation. The outreg2 command in Stata will be used to generate regression tables that provide information for multiple models side-by-side, enabling a comprehensive comparison of results.

The interpretation of the regression results will be context-specific, taking into account the interplay between GDP per capita and other factors (OECD Publishing, 2013b). Potential mechanisms, such as the role of press freedom, will be examined to understand how they contribute to the relationship between income and happiness. The report will emphasize the importance of considering multiple specifications to capture the nuances of this complex relationship.

Conclusion

The findings of the analysis will be summarized, highlighting the key insights obtained from the regression models. The limitations of the analysis will be acknowledged, including potential endogeneity issues, omitted variable bias, and data limitations. Suggestions for future research will be provided, encouraging further exploration of the factors influencing subjective well-being. By addressing the research question and considering its implications, this report aims to contribute to the understanding of the complex relationship between GDP per capita and happiness.

Do-File Appendix

A working do file will be included in the appendix, containing all the Stata commands used for data management, regression analysis, and results generation. The do file will be organized and commented to ensure reproducibility and transparency of the analysis.

References

OECD Publishing. (2013, March 20). Output and analysis of subjective well-being measures. OECD Guidelines on Measuring Subjective Well-being – NCBI Bookshelf. https://www.ncbi.nlm.nih.gov/books/NBK189559/ 

OECD Publishing. (2013b, March 20). Output and analysis of subjective well-being measures. OECD Guidelines on Measuring Subjective Well-being – NCBI Bookshelf. https://www.ncbi.nlm.nih.gov/books/NBK189559/ 

Proto, E., & Rustichini, A. (2013). A Reassessment of the Relationship between GDP and Life Satisfaction. A Reassessment of the Relationship Between GDP and Life Satisfaction, 8(11), e79358. https://doi.org/10.1371/journal.pone.0079358