There are four sections in this test: A, B, C and D. Answer 4 questions.
At
least one question has to be answered from each of the sections A, B and C.
The fourth question can be answered from any section.
Group A
1. Suppose a producer uses capital and labour as inputs. Let capital be
denoted by k, and labour by l. Suppose the production function is y(k,l).
Let w be the wage rate of labour and r be the rental rate of capital.
(a) Provide an example of a production function y which is continuous
everywhere in k and l, yet not differentiable everywhere in k and l,
such that capital and labor are substitutes in production. Write the
function in algebraic form.
(b) For your example, draw the isoquant map using representative
isoquants. Also represent the isoquant(s) algebraically.
(c) For your example, how does the optimal capital choice depend on w
and r?
[5+10+10=25]
2. Consider an 𝑛-firm industry of a homogeneous product; 𝑛 ≥ 2. Firms
play Cournot. Unit cost of production by each firm is 𝑚 > 0. The market
price for the product is 𝑃 = max{0, 𝐴 − 𝑋}, where 𝑋 is industry output
and 𝐴 is the demand shift parameter; 𝐴 > 𝑚. Now consider the merger of
a subset of firms. Assume that a merger of 𝑘 firms will be profitable if
the merged firm’s profit is at least as large as the sum of profits of 𝑘
independent firms in an 𝑛-firm oligopoly industry; 1 < 𝑘 ≤ 𝑛. Answer
the following.
[Note: if 𝑘 firms merge in an 𝑛-firm Cournot oligopoly, the merged firm’s
( )
profit is: 𝜋(𝑘, 𝑛) = ( )
. Don’t derive payoffs separately.]
(a) Examine whether a merger of two firms (i.e., 𝑘 = 2) in an 𝑛 (≥ 2)
firm industry is always profitable.
(b) Derive the condition, in general, for a profitable merger of 𝑘 firms;
2 ≤ 𝑘 ≤ 𝑛.
(c) If there are exactly 5 firms in the industry (i.e., 𝑛 = 5), how many
firms, at the minimum, should merge to make it a profitable merger?
Give the intuition behind your result.
[6+10+9=25]
1
3. (a) An indivisible object is sold to 3 buyers. If any buyer 𝑖 gets 𝑞 ∈ {0,1}
quantity of the object and makes a payment of 𝑝 , her payoff is (𝑞 𝑣 −
𝑝 ), where 𝑣 is the value of the object for buyer 𝑖. Suppose 𝑣 >𝑣 >
𝑣 . Payment 𝑝 can be positive, negative or zero (some buyers may be
paid or compensated).
Consider the following normal form game. The seller asks each buyer
to place a bid. If (𝑏 , 𝑏 , 𝑏 ) are the bids of the buyers then the highest
bidder wins. Ties are broken in favour of higher indexed buyer: for
instance, if 𝑏 ≥ 𝑏 for all 𝑖 ≠ 1, then buyer 1 wins the object. If bidder i
wins, she pays max 𝑏 . Out of this payment, to every buyer k ∈ {1, 2, 3},
the seller returns (i.e., k is compensated)
min 𝑏 ;
if k is the highest bidder or the second highest bidder. If k is the lowest
bidder, then she is compensated
max 𝑏 .
Here the highest, the second highest, and the lowest bidders are
determined by using the same tie-breaking rule as mentioned before - that
is, ties are broken in favour of the higher indexed bidder. For instance, if
𝑏 = 𝑏 = 𝑏 , then 1 is the highest bidder, 2 is the second highest bidder,
and 3 is the lowest bidder; if 𝑏 = 𝑏 > 𝑏 , then 2 is the highest bidder, 3
is the second-highest bidder, and 1 is the lowest bidder.
Show that bidding her own value is a weakly dominant strategy for
each buyer.
(b) Suppose (𝜎 ∗ , 𝜎 ∗ ) is a Nash equilibrium mixed-strategy profile in a
two-person normal form game. Either prove (with appropriate arguments)
or provide a counterexample to the following statement: if a pure strategy
𝑠 is played with positive probability in 𝜎 ∗ , then 𝑠 is not strictly
dominated.
[15=10=25]
2
Group B
1. Consider an agent who lives for two periods. She earns wage income
only in the first period of her life by supplying her labour in inelastic
manner. The timing of the events is as follows – the agent first receives
the wage income W and then allocates her portfolio before any first
period consumption. The agent has the following utility function at the
beginning of her life:
𝑈 = 𝑥 ln(𝑐 ) + 𝛽(1 − 𝑥) ln(𝑐 );
where β is a parameter with 0 < β < 1, and x is a random variable which
can take two values: 1 with probability p and 0 with probability (1 − p).
The realization of x occurs in the first period after she allocates her
resources but before making any consumption in the first period. The
agent has two technologies to invest in: storage technology and
investment technology. Under storage technology, one unit of investment
of the consumption good at the beginning of the first period secures 𝑅 >
0 units of gross return either later in the first period or in the second
period. Investment technology on the other hand guarantees 𝑅 > 0 gross
return in second period but no return if liquidated early. So, precisely,
storage technology is for liquid assets and investment technology is for
illiquid assets. Let θ represent the fraction of savings put in the storage
technology during allocation at the beginning of the first period.
(a) Write down the expected utility maximization problem of the
agent.
(b) Derive the optimum level of θ, that is, the value of θ that
maximizes expected utility. Under what condition on p is the
value of θ equal to 1?
(c) Does optimal utility increase with 𝑅 > 0? In either case show
your result.
[9+(6+4)+6=25]
3
2. Consider an economy producing a single consumption good, 𝑌. The
aggregate profit-maximizing firm produces this good by the single factor
of production, labour, 𝐿, according to the production function 𝑌 = 𝐿 ,
where 0 < 𝜃 < 1. The representative household has labour income, 𝑤𝐿
(where 𝐿 is the labour supplied by the household, and 𝑤 is the wage rate),
and an initial money holding 𝑀 . Its preferences are given by:
𝑈= 𝐶 − 𝐿 ;
where 𝛾 > 1, 0 < 𝛼 < 1; 𝐶 is aggregate consumption, 𝑀 is the money
holding the household ends up with, 𝑃 is the price of good 𝑌 and the last
term, −𝐿 , denotes the disutility from labour supply. The budget
constraint of the household is given by:
𝑃𝐶 + 𝑀 = 𝑤𝐿 + 𝑀 + 𝜋;
where 𝜋 is the household’s profit income. In equilibrium, 𝑌 = 𝐶 and
𝑀 = 𝑀 . Assume price and wage are both flexible.
(a) Solve for the competitive equilibrium output and employment.
(b) Find the effect on output of a monetary expansion ( i.e., find the
sign of ).
[15+10=25]
3. Consider the Solow-Swan growth model. Assume that the production
function is of the form
𝑌 = 𝐹(𝐾, 𝐴𝐿);
where A is the level of technological progress and grows at the rate 𝑔 >
0, L is population which grows at the rate 𝑛 > 0, K is the capital stock
and Y is GDP. Assume that
𝐾̇ = 𝑠𝑌 − 𝛿𝐾;
where the parameter 𝑠 ∈ [0,1] denotes the savings rate and the parameter
4
𝛿 ∈ [0,1] denotes the depreciation on capital. Define 𝑍 = as the
capital labour ratio in efficiency units.
̇
(a) Derive an expression for .
(b) Recall that 𝐹(. , . ) is linearly homogeneous in both its arguments.
Noting that 𝑓(𝑍) = 𝐹(𝑍, 1) where 𝑓(. ) is the intensive form
production function, show that for any 𝜆 > 1,
( ) ( )
< .
Interpret this inequality in words.
(c) Use bullets (a) and (b) to draw a diagram describing the
dynamics of growth in the Solow-Swan model. Plot Z on the x-
axis and the appropriate functions on the y-axis, to show the
̇
dynamics of growth, , relative to the steady state, 𝑍̅
(d) Let 𝑄 = denote output per worker. Show that on the balanced
growth path,
ln(𝑄 ) = ln(𝐴 𝑓(𝑍̅)) + 𝑔𝑡,
where 𝑍̅ is the steady state value of Z and 𝑄 is the balanced
growth path value of Q. Draw a diagram plotting ln(𝑄 ) on the
y-axis and time (t) on the x-axis with both (i) g > 0 and (ii) g =
0. Describe the growth implications across these two cases in a
couple of sentences. India’s savings rate has fallen during
COVID-19. In a separate diagram, show the effects of a fall in
the savings rate on India’s growth in the transition and the long
run using this diagram (under the assumption that g > 0).
[4+5+8+8=25]
5
Group C
1. An economist is asked to evaluate whether having science, technology,
engineering and math institutions (STEM) in a district of India fosters
demand for science education among high school children living in that
district. She has access to two data sets for the year 2011: (1) a data set that
lists the STEM institutes in each district of India, and (2) a data set that is
based on a simple random sample of children of ages 17 to 18 years in
2011; the data set also contains information on the high school stream
choice (Science, Commerce and Arts) of all the surveyed children who are
enrolled in high school. The data set also provides indicators of socio-
economic characteristics of the households the children belong to. Suppose
that all the children in the surveyed age group are enrolled in high school.
Consider the model
𝑌 = 𝛽 + 𝛽 𝑆𝑇𝐸𝑀 + 𝜀 ; (1)
where 𝑖 denotes a child and 𝑑 denotes a district. 𝑌 is a binary variable that
takes the value 1 if the child has chosen science stream and 0 otherwise;
𝑆𝑇𝐸𝑀 denotes the number of 𝑆𝑇𝐸𝑀 institutions in a district.
(a) If the economist estimates 𝛽 using Ordinary Least Squares (OLS),
when will it be consistent? Explain intuitively why it is likely to be
inconsistent. (Do not restate the technical condition in words.)
(b) In light of the inconsistency, the economist decides to add some
control variables. One set of candidate variables for inclusion are
socio-economic characteristics (denote by 𝑍), which will typically
have an important role to play in high school stream choice. These
variables may have some correlation with the number of STEM
institutes in a district, but this correlation is likely to be low.
Moreover, the economist also collects data on some district level
variables like Urbanization rate, Population, Literacy (denoted by
DIST). These variables will have some small direct effect on the
stream choice of students but are likely to be highly correlated to
the number of STEM institutes in a district. Explain why the
inclusion of DIST is more important for the consistency of the
estimated 𝛽 than the inclusion of Z. Why is it important to still
include Z? (NOTE: we are NOT interested in predicting Y.)
(c) Someone states that since there are differences in state policies
regarding education, one should account for them by including state
level dummy variables. Rewrite the empirical model to account for
such differences across the 28 states of India. Explain how the
inclusion of these dummy variables accounts for the differences
across states.
6
(d) Suppose the economist estimates any of the above specifications.
She feels that she can provide proof that her estimator is consistent
by the following procedure: after estimating the model by OLS, she
regresses the residual of the regression on STEM. She contends that
if the 𝑅 of this regression is 0 then the error term does not co-vary
with STEM. Prove that this is an incorrect argument.
(e) Someone points out that it is getting harder nowadays to classify
high school stream choice into the stated three categories because
students take up many combinations that cut across these traditional
streams. But the survey questionnaires have not been updated to
reflect this. So the surveyed students will be required to choose one
of the three options: hence there will be some mismeasurement of
stream choice. What will be the impact of this on properties of the
estimated 𝛽 ?
[5+10+3+5+2=25]
2. Consider the following time series:
𝑌 = 𝜇 + 𝜀 + 𝜃𝜀 ,
where 𝜀 is a white noise process with mean 0, variance 𝜎 ; and all the
autocorrelations of 𝜀 are zeros.
(a) Examine if the time series 𝑌 is stationary or not.
(b) Derive the condition such that 𝑌 is invertible.
(c) Define 𝑍 =𝑌 + 𝑋, where 𝑋 and 𝑌 are independent for all time
points – past, present and future; 𝑋 ∼ 𝑁(0, 10) and 𝑌 is as defined
above. Find the autocorrelation function of 𝑍 . Examine whether 𝑍
is stationary.
[6+6+13=25]
7
3. A fictional island country named NOWHERELAND shut itself from the
world before a single Covid infection was detected. In January of 2021, it
procured enough doses of a single dose vaccine to vaccinate half its
population. Thereafter no more vaccination was done but the economy was
opened - predictably many cases of Covid-19 came up. In May 2021, the
health ministry conducted a survey of all individuals in the country. The
survey collected information on vaccination status, demographic variables
and an indicator of whether the person worked or not (a binary variable).
The ministry wants to estimate the benefits of vaccination: in particular, it
wants to contend that the vaccine had positive effects on an individual's
labour supply. You are the analyst.
(a) Suppose vaccination centres gave vaccines to people, who came
voluntarily, on a first come first served basis - till they exhausted the
vaccines they had (PROTOCOL 1).
(i) Write down an empirical model, to estimate by Ordinary Least
Squares, the impact of getting a vaccine on an individual's labour
supply. Under what condition(s) will the estimated parameter be
the true impact of vaccination?
(ii) Suppose you discover that the ministry conducted a similar survey
in November 2020 (for obvious reasons the question on vaccine
status was not asked). You therefore have two observations on
each person. Will this help you to come up with a better
specification to estimate the impact of vaccination? If yes, write
down the empirical model and explain why. If not, with the help
of the same empirical model, explain why not.
(b) Suppose vaccination centres now followed a different protocol
(PROTOCOL 2). They allowed vaccines only to people 45 years old and
above. The modality was the same as above – first come first served -
and getting a vaccine shot was voluntary. Can you use this age cut-off to
suggest a method to estimate the impact of the vaccine on labour supply?
You have access to both the surveys mentioned above.
(c) Now suppose that NOWHERELAND had followed a completely
different protocol. It had listed down all the households and used a
random number generator to draw up the list of the individuals (half the
population) that should be vaccinated (PROTOCOL 3). The individuals
eligible for the vaccine (in the list) could however choose not to take it.
Using the same surveys how would you estimate the impact of
vaccination on labour supply? Write down the empirical model and
describe how estimation under this protocol may lead to better estimates
than in the previous two protocols.
[(5+8)+6+6=25]
8
Group D
1. Carbon dioxide emissions vary across countries and across individuals
within countries.
(a) Which countries are the three largest emitters of CO2 per year?
(b) What are the main determinants of annual CO2 emissions of
countries? What about emissions per capita? Explain the reasons
why the factors you have listed explain cross-country variation in
CO2 emissions per capita.
(c) Which variable accounts the most for variation in CO2 emissions
across individuals within countries?
(d) What policies do you think governments across the world, and in
India in particular, could adopt to reduce CO2 emissions, or at least
reduce increases in CO2 emissions, cost-effectively?
(e) How could these policies be designed or modified to be pro-poor
while still being cost-effective?
[3+8+1+8+5=25]
2. (a) What was the share of health expenditure in the GDP of India before the
recent pandemic? What percentage of health spending was public (i.e.,
by government?)
(b) How do these compare to the global averages and to other lower-middle
income countries?
(c) What are some of the reasons why the share of health expenditure in
India compares to the global average in the way that it does?
(d) Some high-income countries have almost complete public provision of
health care – like the UK, some have mostly private health providers
but nationalized health insurance, so that all health services are paid for,
but not provided by, the government (as in Canada); and some have
private health care and private health insurance but with regulations and
financial support to ensure that everyone is covered (Germany and
Switzerland). How can good quality and sufficient health care be
provided to everyone in India? Explain how you think India should
move towards this goal and how it can be financed. Provide reasons
why you think this is the best path forward.
[2+4+9+10=25]
