Faculty of Arts and Social Sciences
School of Economics
ECOS3002 Development Economics
The university exams office is responsible for administering the exam, through the special ‘In-semester Test for:
ECOS3002’ Canvas site.
All official information on the exam and its administration, comes from them, and overrides anything I say in this
video or elsewhere on the ECOS3002 Canvas site.
I will not be available on the day of the exam, the Ed site will be offline the day of the exam, and exams cannot be
submitted by email.
Any exam-related issues will have to be dealt with through official University systems (e.g., Special Consideration), and
approval of appeals is not guaranteed.
This video is providing an informal review of the logistics of the exam, and a review of some of the relevant
content on the exam.
I highly recommend logging in to the exam site as soon as you have access, reading through everything, and
making sure you have access to all the materials, resources, and software necessary for an online exam.
Exam details
Date of test: 11/10/2021 (Monday)
Start: 14:00 AEST
Duration: 1 hours and 30 minutes (90 minutes). This includes:
10 minutes reading time, but you are free to start the test as soon as you are ready.
30 minutes of upload time to allow you to upload your files as per your test instructions. Do NOT treat this as
extra writing time. The upload time must be used solely to save and upload your files correctly as per the test
instructions. Manage your time carefully. Check that you have saved and named your file correctly and uploaded
the correct file. If your time runs out while you are uploading this is not considered a technical issue.
Materials required: (i) scientific calculator, and (ii) a sheet of blank paper with a writing instrument (pen or
pencil), OR a digital drawing tool.
Your final exam submissions will be in the form of a pdf (only).
Analysis
The exam will not involve complex calculations or manipulations in Excel, however it will
involve basic operations that you can implement on a scientific calculator.
You will also need to create a figure – you can do that using pen/pencil and paper, or a
digital drawing tool. Either way you will need to upload your figure as a pdf.
Exam format
Question type Points Recommended time spent
Question 1 Draw, calculate, interpret 15 15 minutes
Question 2 Short answer: interpret a quasi-experiment 10 10 minutes
Question 3 & 4 Short answer 5 each 5 minutes each
Question 5 Short essay 15 15 minutes
Academic honesty
It should go without saying that the exam is to be taken completely individually. Use of
any method to communicate with classmates during the exam is forbidden.
Beyond that, it is an open book exam. The exam is designed so that you won’t get a huge
benefit from searching online or in your textbook, so don’t get tempted to plan to just
look things up for your exam responses. But you are certainly welcome to use either to
look up concepts, definitions, etc.
Faculty of Arts and Social Sciences
School of Economics
ECOS3002 Development Economics
Mid-sem exam review
Content overview
Content of exam
Everything up to and including week 7 is fair game: lectures, tutorials, and textbook
chapters.
In practice we the exam is most heavily focused through week 6, with light coverage of week 7
(enough to review the lecture video).
Week Week Beginning Lecture Lecture Topic(s) / textbook chapter(s)
1 9 Aug Lecture 1 Chapter 1: What is development? Indicators and issuesChapter 4 (part 1): Impact evaluation
2 16 Aug Lecture 2 Chapter 4 (part 2): Impact evaluationChapter 3: History of thought in development economics
3 23 Aug Lecture 3 Chapter 5: Poverty and vulnerability analysisChapter 6: Inequality and inequity
4 30 Aug Lecture 4 Chapter 10: The economics of farm households
5 6 Sept Lecture 5 Chapter 18: Agriculture for development
6 13 Sept Lecture 6
Chapter 11: Population and development
Chapter 12: Labour and migration
Chowdhury research vignette
7 20 Sept Lecture 7 Chapter 13: Financial services for the poor
Chapter 1: What is development?
Indicators and issues
The first question to answer about development is – what is it? How do we define it? How do we quantify it?
Our textbook posits 7 dimensions of development:
1. Income and income growth: totals like GDP, GNP, GNI, per capita, growth rate, PPP conversion.
2. Poverty and hunger: % below a poverty line (monetary), or a metric like calories.
3. Inequality and inequity: comparing top X% vs bottom Y%; inequity about opportunities.
4. Vulnerability: risk of poverty, vulnerability or susceptibility to adverse shocks (covariate and idiosyncratic risk),
poverty traps?
5. Basic needs: human development: human capital (health, education), HDI, multidimensional poverty indices.
6. Environmental sustainability: intergenerational equity.
7. Quality of life: many broader theories, some outside economics. Within economics, Sen’s capabilities approach
(focused on what you could do – freedom of choice), and Easterly’s 81 indicators of quality of life.
Chapter 1: What is development?
Indicators and issues
An approach to quantify well-being is through subjective measures, like “subjective well
being” or happiness.
Provides a single-index measure, all encompassing, going beyond money.
But how well can we measure it? Easterlin paradox (1974), showing no correlation between
income and happiness in OECD, seems to be overturned in developing countries (e.g., Deaton,
2008).
The dominant international development framework is the Sustainable Development
Goals (declared in 2016 with targets for 2030) to replace the Millenium Development
Goals (2000).
Chapter 4: Introduction to impact
evaluation and RCTs
An important trend in international development is to evaluate the effectiveness of
international development programs and policies, using causal inference techniques.
The challenge for an impact evaluation researcher is that without a research design,
data on programs and policies is almost always suffers from selection bias – because
there are choices (on demand side or supply side of an intervention) about whether or
not to take up an intervention, the take-up decision can be affected by hard-to-measure
characteristics that also affect outcomes.
Then if outcomes differ between recipients and non-recipients of an intervention, was it
because of those characteristics (which we can’t measure and control for), or the intervention?
Impact evaluation methods provide causal inference techniques to help us overcome selection
bias.
Chapter 4: Introduction to impact
evaluation and RCTs
The randomized control trial (RCT) is considered the most rigorous or most scientific method to
overcome selection bias. It is based on the clearest research design, with the weakest
assumptions.
Because we explicitly randomize participants into treatment and control groups, we control the
allocation of treatment, so treatment allocation shouldn’t be correlated with hard-to-measure
characteristics.
Even here, whether “randomization worked” on unobservables is untestable, however we do balance
checks on observables to verify.
With an up-front research design, RCTs lead to clear, simple analysis. Two common methods to
estimate effects from RCTs are ITT (an average treatment effect) and ToT (a local average
treatment effect).
Chapter 4: Introduction to impact
evaluation and RCTs
Because of randomization, RCTs are highly internally valid. But they may suffer from external
validity issues, especially if we work with an opportunistic sample (e.g., a single NGO or
company). This can also cause a pioneer effect.
Because RCTs are heavily controlled/planned, they can be subject to common experimental
biases – e.g., Hawthorne effect (being studied changes behavior), John Henry effect (control
group tries to catch up).
RCTs rely on the SUTVA assumption. Sometimes we may need to randomize a larger scale (e.g.,
village / neighborhood) to mitigate spillovers.
In some cases we can leverage “natural” randomization (e.g., that a government implemented).
There we want to particularly check that randomization worked.
Chapter 4: Introduction to impact
evaluation and RCTs
There are other credible ways to do an impact evaluation.
In economics these are known as “quasi-experimental” methods because they try to
imitate what a pure experiment does – separating treatment from the characteristics of
the treated units.
Common methods in applied economics include:
Regression discontinuity design (RDD)
Differences-in-differences (DiD)
Instrumental variables (IV)
Propensity score matching (PSM)
Chapter 4: Introduction to impact
evaluation and RCTs
While we can learn a lot from these methods, they all suffer drawbacks relative to RCTs:
RDD only estimates a local average treatment effect (LATE), though we model the treatment
allocation process.
DiD relies on assumptions about unobservable counterfactual trends.
IV relies on an untestable assumption, the exclusion restriction, and again only gives us a
LATE, typically for an undefined population.
PSM relies on strong assumptions around how observables allow us to balance
unobservables.
What are the threats to validity of these quasi-experimental designs, and how would
you test for them?
Chapter 3: History of thought in
development economics (post-WWII)
1950s-1960s: “glory years” of recovery, big push theories used to drive recovery in Europe.
1970-1982: growth boom in 50s-60s didn’t lead to poverty reduction. Development agenda
expanded beyond pure growth, to look at pro-poor growth and other dimensions of
development. Lots of fiscal spending and debt accumulation. 1970s were also a major
inflationary period.
1982-1997. Era starts with debt crises, as high inflation means high and unsustainable interest
rates. To combat this, we get structural adjustment reforms under so-called Washington
consensus, which was about opening up markets and reducing the role of the state in markets
(deregulation, privatization, lowering of trade barriers, etc).
Chapter 3: History of thought in
development economics (post-WWII)
However Washington consensus was too abrupt a change, many countries couldn’t adapt. 1990s
considered a “lost decade” for development, particularly in Africa.
1997-2019. As a corrective, renewed role of the state in complementing the market,
multidimensionality in development, more customized development policies. Emergence of MDG
agenda (2000) with eye to 2015.
End of cold war (1989) brings greater interest in aid performance, and emergence of the impact
evaluation revolution in development economics, in parallel to the credibility revolution in
economics. Key leaders such as Abhijit Banerjee, Esther Duflo, Michael Kremer (Nobel Prize,
2019).
Chapter 5: Poverty and vulnerability
analysis
Poverty means not having a sufficient amount and/or quality of something. Measurement then
involves defining that amount/quality, and then identifying which individuals/households don’t
have a sufficient amount/quality.
Typically we want a monetary measure. While we might like to use income, in practice we
typically use consumption (expenditure), adjusted for, e.g., CPI (inflation over time), PPP, access
to public goods, converted to per capita level.
Set a poverty line – e.g., extreme poverty line (enough money for required daily caloric intake),
normal poverty line (often 2x extreme poverty line), international poverty line (PPP$1.90 per day
for extreme poverty, and PPP$3.10 per day for normal poverty).