LAISHA
Open Laisha

Indian Statistical Service Examination Causal Inference, Panel Data & Efficiency Analysis: practice questions

Indian Statistical Service Examination (with Indian Economic Service) · India · Union Public Service Commission

Original practice questions written to the published syllabus · not past papers
10worked here, free
59written to this exam in this subject
8syllabus units

About Indian Statistical Service Examination

Indian Statistical Service Examination (with Indian Economic Service) is a government recruitment exam run by Union Public Service Commission, for candidates in India. Written examination of six papers and a viva voce. Reported scheme: General English 100, General Studies 100, Statistics I and II (objective, 200 each), Statistics III and IV (descriptive, 200 each); viva 200.

Official site: upsc.gov.in.

We have not been able to confirm these details against the exam body's own specification, so fees, dates and pass marks are left out on purpose. Check the current format, eligibility, fees and dates on the official site.

Causal Inference, Panel Data & Efficiency Analysis in Indian Statistical Service Examination

Did the scheme, technology or variety cause the change? Potential outcomes, matching and weighting, difference-in-differences, instrumental variables, regression discontinuity and synthetic control, panel data models, quantile regression, and stochastic frontier and DEA efficiency analysis. Classical econometrics (OLS violations, simultaneous equations) is in the Agribusiness track and is assumed here.

Our 59 Indian Statistical Service Examination questions in this subject: 18 easy, 27 medium, 14 hard, across 8 units.

10 practice questions, worked

Choose an answer, then open the working. Each explanation says why the right answer is right and why the others are not.

These 10 were written to Indian Statistical Service Examination and are not shown on any other exam's page.

Question 1 · easy

Why can the causal effect of a scheme on an individual farm never be observed directly?

  1. Only one potential outcome is seen
  2. Schemes affect every farm by exactly the same amount
  3. Farm outcomes are always measured with large error
  4. Individual effects are always zero on average
Answer and working

A. Only one potential outcome is seen

Each farm is either treated or not, so only Y(1) or Y(0) is observed; the other is the counterfactual. This is the fundamental problem of causal inference, so averages are estimated instead.

Question 2 · medium

A voluntary drip-irrigation subsidy is evaluated. Which effect is usually the most policy-relevant for judging the scheme as it ran?

  1. The average effect on the untreated (ATU)
  2. The correlation between subsidy and yield
  3. The average effect on the treated (ATT)
  4. The effect on the single largest farm
Answer and working

C. The average effect on the treated (ATT)

The ATT measures what the scheme did for those who took it; the ATU concerns extending it to non-participants; a correlation is not a causal effect.

Question 3 · hard

In a causal graph, both adopting a technology and high yield cause a farm to be selected for a media award. Analysing only award-winning farms will:

  1. Turn the award into a valid instrument
  2. It conditions on a collider
  3. Leave the adoption-yield relation unchanged
  4. Remove all confounding between adoption and yield
Answer and working

B. It conditions on a collider

The award is a common effect (collider) of adoption and yield; conditioning on it induces an association between them even if none exists in the population.

Question 4 · easy

A first-stage F-statistic for the excluded instrument is 4. What does this indicate?

  1. Overidentification has been rejected
  2. The exclusion restriction holds
  3. A weak instrument; IV is unreliable
  4. A strong instrument; IV is preferred to OLS
Answer and working

C. A weak instrument; IV is unreliable

F below 10 (Staiger and Stock rule of thumb) signals a weak instrument: IV is biased towards OLS and inference is unreliable; weak-instrument-robust tests such as Anderson-Rubin are needed.

Question 5 · medium

Many farmers report landholdings of exactly 1.98 to 2.00 hectares, just inside a 2-hectare eligibility limit. What does this suggest?

  1. A larger treatment effect at the cut-off
  2. That the design is sharp
  3. Manipulation of the running variable
  4. That covariates are balanced
Answer and working

C. Manipulation of the running variable

Bunching just on the favourable side indicates sorting, which a McCrary-type density test detects; it undermines the comparison.

Question 6 · easy

To whom does an RD estimate apply?

  1. Units near the cut-off
  2. All units in the population equally
  3. Only the units with the largest scores
  4. Only units far from the cut-off
Answer and working

A. Units near the cut-off

RD identifies the effect at the threshold; extrapolating to farms far from it needs further assumptions.

Question 7 · medium

A synthetic control tracks the treated state poorly before the policy, with a large pre-treatment RMSPE. What follows?

  1. The gap is not a credible effect
  2. The method is valid if donors are many
  3. The policy had no effect
  4. The effect is larger than the gap
Answer and working

A. The gap is not a credible effect

A credible counterfactual must reproduce the pre-treatment path; with poor fit the gap may reflect mismatch rather than the policy.

Question 8 · hard

In a dynamic panel, pooled OLS gives rho = 0.85 and fixed effects gives rho = 0.45. Where is the true rho likely to lie?

  1. Exactly at 0.65
  2. Between 0.45 and 0.85
  3. Above 0.85
  4. Below 0.45
Answer and working

B. Between 0.45 and 0.85

Pooled OLS is biased upward (lag correlated with the unit effect) and FE downward (Nickell bias), so the estimates usually bracket the true value; GMM estimates should fall within the bracket.

Question 9 · medium

In Arellano-Bond difference GMM, what serves as instruments for the differenced lagged outcome?

  1. The current level of the outcome
  2. The differenced error term
  3. Lagged levels from t - 2 and earlier
  4. Future values of the outcome
Answer and working

C. Lagged levels from t - 2 and earlier

If u(it) is not serially correlated, y(i,t-2) and earlier are uncorrelated with the differenced error and correlated with the differenced lag.

Question 10 · medium

Why does instrument proliferation in GMM cause concern?

  1. It removes all bias from the estimates
  2. It overfits and weakens the Hansen J test
  3. It only affects random-effects models
  4. It makes the estimates exactly identified
Answer and working

B. It overfits and weakens the Hansen J test

Many instruments overfit the endogenous variables and make the Hansen test implausibly accept, so the instrument count should be reported and limited.

A study plan for Causal Inference, Panel Data & Efficiency Analysis in Indian Statistical Service Examination

This plan follows where our questions actually fall, so the time goes where the marks are likely to be. It is our reading of the syllabus, not the exam body's weighting.

  1. First, the heavy units (41% of the 59 questions): Difference-in-differences and event studies (8, 14%); Dynamic panels (introduction to GMM) (8, 14%); Instrumental variables and local average treatment effects (8, 14%).

    Read the manual chapters for these first and do every worked question you can find on them.

  2. Then the middle band: Panel data: fixed and random effects and the Hausman test (8, 14%); Potential outcomes and the identification problem (7, 12%); Quantile regression (7, 12%).

    Aim for one sitting per unit, then mixed questions across them.

  3. Then the lighter units: Regression discontinuity designs (7, 12%); Synthetic control (6, 10%).

    Learn the definitions, classifications and standard numbers; these units are more often recalled than worked.

  4. Leave time for the hard ones: Dynamic panels (introduction to GMM) (3 hard of 8), Potential outcomes and the identification problem (2 hard of 7), Difference-in-differences and event studies (2 hard of 8).

    Come back to these after a first pass, untimed, and read every explanation, including for the options you rejected.

  5. Bank the easy marks: Panel data: fixed and random effects and the Hausman test (5 easy of 8); Synthetic control (3 easy of 6).

    Do these early in revision so they are fast on the day.

  6. Finish timed. Sit the subject as a timed set without notes, mark it, and send each wrong answer back to its unit above.

Last updated · How Laisha writes and checks its pages · Report a mistake

Ask Laisha about this

Opens Laisha's chat with your question ready to send. Sign in or start free.

Unlock every question — Indian Statistical Service Examination Causal Inference, Panel Data & Efficiency Analysis

US$10 a month · one price worldwide · cancel any time

  • Every practice question in this subject, with full working
  • Timed exam papers and progress by unit
  • Ask Laisha to explain any answer — up to 1,000 questions a month
Free or Smarter
FreeSmarter
Grow guides, pest pages, articlesRead allRead all
Ask Laisha5 a dayUp to 1,000 a month
Exam practiceThe sample on each pageEvery question, timed papers
Crop calendar and pest alerts—For your own fields
Calculators and tools—All
Uploads and reports—100 a month, real files
10 days free for every new account, no card.