LAISHA
Open Laisha

RBI Grade B Time Series 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
74written to this exam in this subject
9syllabus 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.

Time Series Analysis in RBI Grade B

ARIMA modelling, unit-root tests, ARCH/GARCH, VAR and cointegration, and forecast evaluation for economic and financial series.

Modelling observations ordered in time, from classical decomposition to Box-Jenkins and beyond: stationarity, ARIMA and seasonal ARIMA, volatility models for prices, vector autoregressions, state-space models and the Kalman filter, and forecast evaluation. Agribusiness econometrics introduces ARIMA and unit roots; this subject is the full treatment.

Our 74 RBI Grade B questions in this subject: 23 easy, 41 medium, 10 hard, across 9 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

For which process does simple exponential smoothing give the minimum mean squared error one-step forecast?

  1. White noise around a fixed linear trend
  2. ARIMA(0,1,1)
  3. A stationary AR(1) process with a large positive coefficient
  4. ARIMA(0,2,2), a doubly integrated process
Answer and working

B. ARIMA(0,1,1)

SES is the optimal forecast for ARIMA(0,1,1), a random walk observed with noise. Holt's linear method corresponds to ARIMA(0,2,2); a stationary AR(1) and a trend plus noise need different forecasts.

Question 2 · medium

An analyst models log prices with ARIMA and reports the antilog of the log-scale point forecast as the expected price. What is wrong?

  1. The antilog of a forecast always overstates the mean price
  2. The antilog estimates the median, not the mean, price
  3. Log models cannot be used for prices that are always positive
  4. ARIMA models cannot be fitted to logged data at all
Answer and working

B. The antilog estimates the median, not the mean, price

If log price is normal, exp(forecast) is the median; the mean is larger, approximately exp(forecast + variance/2). The antilog therefore understates, not overstates, the mean. Logs are commonly used for prices.

Question 3 · medium

The residual ACF of a fitted non-seasonal ARIMA model for monthly data is inside the limits at all lags except a large spike at lag 12. What is the most sensible next step?

  1. Accept the model, since one spike is expected by chance
  2. Take a second ordinary difference of the series
  3. Increase the non-seasonal AR order to 12
  4. Add a seasonal term at lag 12
Answer and working

D. Add a seasonal term at lag 12

A large residual spike at the seasonal lag in monthly data signals unmodelled seasonality; a seasonal MA or AR term (or seasonal differencing) is needed. A full AR(12) wastes parameters, and a second difference addresses trend, not seasonality.

Question 4 · medium

An ADF test with a constant but no trend gives a statistic of -2.50. The 5 per cent critical value is -2.86. What is the conclusion?

  1. Reject the unit root, because -2.50 is below -1.96
  2. Reject the unit root, because -2.50 lies above -2.86
  3. The test is invalid because the statistic is negative
  4. Do not reject the unit root at 5 per cent
Answer and working

D. Do not reject the unit root at 5 per cent

The null is rejected only if the statistic is more negative than the critical value. -2.50 is not below -2.86, so the unit root is not rejected. Using the normal value -1.96 is the classic error: the DF statistic does not follow the t or normal distribution.

Question 5 · medium

A GARCH(1,1) model for weekly returns has omega = 0.2, alpha = 0.10 and beta = 0.85. What is the unconditional variance of returns?

  1. 1.33
  2. 0.211
  3. 4.0
  4. 0.222
Answer and working

C. 4.0

Unconditional variance = omega/(1 - alpha - beta) = 0.2/0.05 = 4.0. Dividing by 1 - beta only gives 1.33; dividing by alpha + beta gives 0.211.

Question 6 · easy

Under what condition is a GARCH(1,1) process covariance stationary?

  1. alpha + beta < 1
  2. alpha < beta
  3. alpha + beta = 1
  4. omega < alpha + beta
Answer and working

A. alpha + beta < 1

With alpha + beta < 1 the unconditional variance omega/(1 - alpha - beta) is finite. alpha + beta = 1 is integrated GARCH, where variance shocks never die out.

Question 7 · medium

Engle's LM test regresses squared residuals from a price-return model on 4 of their own lags, with n = 200 and R-squared = 0.06. The 5 per cent critical value of chi-square with 4 df is 9.49. What is the conclusion?

  1. LM = 12; ARCH effects are present
  2. LM = 12; but the test needs an F table, not chi-square
  3. LM = 3; no ARCH effects at 5 per cent
  4. LM = 6; no ARCH effects at 5 per cent
Answer and working

A. LM = 12; ARCH effects are present

LM = n R-squared = 200 x 0.06 = 12 > 9.49, so the null of no ARCH is rejected. Dividing by the number of lags or using R-squared x 100 gives 3 or 6, both below the critical value, and wrongly accepts the null.

Question 8 · medium

What is meant by volatility clustering in agricultural price returns?

  1. Large changes tend to follow large changes, and small follow small
  2. Returns are concentrated in a few months of the marketing year
  3. Price changes have a strong positive autocorrelation at lag 1
  4. Prices of related crops tend to move together in the same direction
Answer and working

A. Large changes tend to follow large changes, and small follow small

Volatility clustering means the size (not the sign) of changes is autocorrelated, seen as autocorrelation in squared returns. Co-movement of crops is correlation across series, and returns themselves usually show little autocorrelation.

Question 9 · easy

Monthly prices (actual) were 100, 80, 120 and 50; the forecasts were 90, 88, 108 and 60. What is the MAPE?

  1. 10.0 per cent
  2. 12.0 per cent
  3. 12.5 per cent
  4. 11.4 per cent
Answer and working

C. 12.5 per cent

Absolute percentage errors are 10/100, 8/80, 12/120 and 10/50 = 10%, 10%, 10% and 20%, mean 12.5%. Dividing by the forecasts gives 12.0%; total absolute error over total actuals gives 11.4%, a weighted measure, not MAPE; 10.0 is the MAE in price units, not a percentage.

Question 10 · easy

Why is MAPE unsuitable for evaluating forecasts of weekly arrivals at a small market that has some weeks with zero arrivals?

  1. It is undefined when the actual value is zero
  2. It always favours models with a trend term
  3. It cannot be computed for weekly data at all
  4. It is measured in the same units as the data
Answer and working

A. It is undefined when the actual value is zero

MAPE divides each error by the actual value, so zero actuals make it undefined and near-zero actuals make it explode. Scaled errors such as MASE avoid this.

A study plan for Time Series 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 (38% of the 74 questions): Forecast accuracy: MAPE, RMSE and the Diebold-Mariano test (10, 14%); Box-Jenkins ARIMA: identification, estimation and diagnostic checking (9, 12%); Unit-root tests and structural breaks (9, 12%).

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

  2. Then the middle band: Vector autoregression and cointegration (9, 12%); AR, MA and ARMA processes (8, 11%); ARCH and GARCH models for price volatility (8, 11%).

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

  3. Then the lighter units: Seasonal ARIMA and ARIMAX with weather covariates (8, 11%); Stationarity, autocorrelation and partial autocorrelation functions (8, 11%); Smoothing: moving averages and exponential smoothing (Holt-Winters) (5, 7%).

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

  4. Leave time for the hard ones: Seasonal ARIMA and ARIMAX with weather covariates (2 hard of 8), Box-Jenkins ARIMA: identification, estimation and diagnostic checking (2 hard of 9), Unit-root tests and structural breaks (2 hard of 9).

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

  5. Bank the easy marks: Stationarity, autocorrelation and partial autocorrelation functions (4 easy of 8); AR, MA and ARMA processes (3 easy of 8).

    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 Time Series 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.