LAISHA
Open Laisha

RBI Grade B Causal Inference, Panel Data & Efficiency Analysis: practice questions

RBI Grade B (DSIM) - Department of Statistics and Information Management · India · Reserve Bank of India Services Board

Original practice questions written to the published syllabus · not past papers
10worked here, free
77written to this exam in this subject
10syllabus units

About RBI Grade B

RBI Grade B (DSIM) - Department of Statistics and Information Management is a government recruitment exam run by Reserve Bank of India Services Board, for candidates in India. Phase I objective screening, Phase II papers in statistics and English, and an interview (reported).

Official site: opportunities.rbi.org.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 RBI Grade B

Econometric identification: endogeneity and instrumental variables, panel data fixed and random effects with the Hausman test, and basic programme-evaluation designs.

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 77 RBI Grade B questions in this subject: 21 easy, 40 medium, 16 hard, across 10 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 RBI Grade B and are not shown on any other exam's page.

Question 1 · medium

After matching, mean landholding is 5.2 acres among treated and 4.8 acres among matched controls, with a pooled standard deviation of 2.0 acres. What is the standardised mean difference, and is balance adequate by the 0.1 rule of thumb?

  1. 0.40; adequate
  2. 0.20; adequate
  3. 0.083; adequate
  4. 0.20; not adequate
Answer and working

D. 0.20; not adequate

SMD = (5.2 - 4.8)/2.0 = 0.20, above 0.1, so landholding is still imbalanced. 0.40 is the raw difference; dividing by the control mean gives a percentage difference, not an SMD.

Question 2 · medium

In an event study, coefficients for the three years before a scheme are large and significant. What does this suggest?

  1. Pre-trends or anticipation
  2. That the scheme was randomly assigned
  3. That standard errors must be too large
  4. A large and genuine treatment effect
Answer and working

A. Pre-trends or anticipation

Leads should be near zero under parallel trends; significant leads cast doubt on the design.

Question 3 · medium

A scheme covers only paddy in treated districts. Comparing paddy and other crops within treated and untreated districts over time is called:

  1. A synthetic control design
  2. A regression discontinuity design
  3. A triple-differences design
  4. An instrumental-variables design
Answer and working

C. A triple-differences design

DDD adds a third comparison (crop) to net out district-specific trends affecting all crops, relaxing the parallel-trends assumption.

Question 4 · hard

A DiD study has only six treated and six control states. Which inference approach is recommended?

  1. Wild cluster bootstrap
  2. Conventional robust standard errors with no clustering
  3. Clustering at the farm level instead
  4. Using the normal distribution with 12 clusters
Answer and working

A. Wild cluster bootstrap

With few clusters, cluster-robust standard errors are biased downward; the wild cluster bootstrap gives more reliable inference.

Question 5 · hard

With three instruments for one endogenous variable, the Hansen J test does not reject. What can be concluded?

  1. All three instruments are certainly valid
  2. They agree; validity is unproven
  3. The instruments are weak
  4. The model is exactly identified
Answer and working

B. They agree; validity is unproven

Overidentification tests check whether instruments give similar estimates, so they are mutually consistent but not proven valid; if all are invalid in the same way, the test cannot detect it.

Question 6 · medium

In placebo tests, the treated state's post/pre RMSPE ratio is the largest among 20 states. What is the permutation p-value?

  1. 0.01
  2. 0.05
  3. 0.95
  4. 0.20
Answer and working

B. 0.05

Ranked first of 20 units, the p-value is 1/20 = 0.05.

Question 7 · hard

One state's export ban diverts grain trade to neighbouring states in the donor pool. What problem arises?

  1. The treated state leaves the sample
  2. The pre-treatment fit becomes perfect
  3. Spillovers contaminate the synthetic control
  4. Weights may no longer sum to one
Answer and working

C. Spillovers contaminate the synthetic control

Donors affected by the treatment no longer represent the untreated counterfactual; affected donors should be excluded.

Question 8 · easy

What does the Breusch-Pagan Lagrange multiplier test decide in panel data?

  1. Random effects versus pooled OLS
  2. Normal versus non-normal errors
  3. Fixed effects versus random effects
  4. Static versus dynamic panels
Answer and working

A. Random effects versus pooled OLS

Its null is no unit-specific variance (pooled OLS adequate); the Hausman test compares FE and RE.

Question 9 · medium

When is system GMM preferred to difference GMM?

  1. When the panel has only two periods
  2. When there is no unit effect
  3. When all regressors are strictly exogenous
  4. When the series is highly persistent
Answer and working

D. When the series is highly persistent

With rho near 1, lagged levels weakly predict differences; system GMM adds level equations instrumented by lagged differences.

Question 10 · hard

A treatment's conditional quantile effect at tau = 0.1 is large. Does this mean the farms that were poorest before treatment gained most?

  1. No; ranks need not be preserved
  2. Yes, by definition of quantile effects
  3. Yes, if the sample is large
  4. No, quantile effects are always zero
Answer and working

A. No; ranks need not be preserved

Quantile effects compare distributions, not individuals; without rank invariance, the farms at the 10th percentile after treatment need not be those that were there before.

A study plan for Causal Inference, Panel Data & Efficiency Analysis in RBI Grade B

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 (44% of the 77 questions): Propensity score matching and weighting (9, 12%); Randomised evaluation in agriculture and its limits (9, 12%); Difference-in-differences and event studies (8, 10%); Dynamic panels (introduction to GMM) (8, 10%).

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

  2. Then the middle band: Instrumental variables and local average treatment effects (8, 10%); Panel data: fixed and random effects and the Hausman test (8, 10%); Potential outcomes and the identification problem (7, 9%); Quantile regression (7, 9%).

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

  3. Then the lighter units: Regression discontinuity designs (7, 9%); Synthetic control (6, 8%).

    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 — RBI Grade B 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.