LAISHA
Open Laisha

UGC-NET Econometrics: practice questions

UGC-NET (June 2025) · India · National Testing Agency

Original practice questions written to the published syllabus · not past papers
10worked here, free
88written to this exam in this subject
15syllabus units

About UGC-NET

UGC-NET (June 2025) is a PhD, fellowship and lectureship exam run by National Testing Agency, for candidates in India. CBT (since Dec 2018). Paper I 50 questions / 100 marks; Paper II 100 questions / 200 marks; 180 minutes total; +2, no negative marking.

Source: ugcnet.nta.ac.in.

Checked against the exam body's published information. Formats and fees change; confirm the current edition there before you book.

Econometrics in UGC-NET

Properties of OLS estimators, Gauss-Markov, ML, violations and tests, simultaneous-equation identification and estimation, stationarity and unit roots, binary choice. Largely conceptual MCQs with occasional short computations.

Estimating economic relationships from data and knowing when the estimates can be trusted: OLS and its assumptions, their violations, simultaneous equations, time series and binary-choice models.

Our 88 UGC-NET questions in this subject: 12 easy, 48 medium, 28 hard, across 15 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 UGC-NET and are not shown on any other exam's page.

Question 1 · medium

To test constant returns and equal labour and capital elasticities (2 restrictions) in a production function with 5 coefficients on 33 farms, RSS is 150 restricted and 120 unrestricted. The F statistic is

  1. 7.00
  2. 4.12
  3. 3.50
  4. 2.80
Answer and working

C. 3.50

F = [(RSS_R - RSS_U)/m]/[RSS_U/(n - k)] = (30/2)/(120/28) = 3.50 with (2, 28) df. Forgetting to divide by m gives 7.00; using n instead of n - k gives 4.12; the denominator must use the unrestricted RSS (150 gives 2.80).

Question 2 · medium

Under heteroscedasticity, if the conventional OLS standard errors are used for inference,

  1. the t and F tests may be misleading because the variance formula is wrong
  2. the coefficient estimates usually become biased
  3. R-squared becomes negative
  4. the tests remain exactly valid because OLS is still unbiased and consistent under it
Answer and working

A. the t and F tests may be misleading because the variance formula is wrong

The usual formula assumes a constant error variance, so it misstates the true sampling variance; t and F statistics then do not follow their nominal distributions. The coefficients themselves stay unbiased.

Question 3 · hard

For a Goldfeld-Quandt test, 40 farms are ordered by size and 8 central observations dropped. Separate regressions with 3 coefficients give RSS = 36 (small farms) and 117 (large farms). The test statistic and its degrees of freedom are

  1. F = 3.25 with (13, 13) df
  2. F = 3.25 with (16, 16) df
  3. F = 0.31 with (13, 13) df
  4. F = 3.25 with (15, 15) df
Answer and working

A. F = 3.25 with (13, 13) df

Each sub-sample has (n - c)/2 = 16 observations, and each regression uses k = 3 coefficients, so df = (n - c - 2k)/2 = (40 - 8 - 6)/2 = 13. F = RSS2/RSS1 with the larger-variance group on top = 117/36 = 3.25. Omitting the k deduction gives 16 df; putting the smaller RSS on top gives 0.31. 15 df deducts only one parameter per sub-sample.

Question 4 · hard

In a Breusch-Pagan-Godfrey test with three Z variables, the regression of scaled squared residuals on the Z's has explained sum of squares 14.6. The 5 percent chi-square critical value with 3 df is 7.81. The conclusion is

  1. statistic = 14.6; reject homoscedasticity
  2. statistic = 7.3; do not reject homoscedasticity
  3. statistic = 4.87; do not reject homoscedasticity
  4. statistic = 7.3; reject homoscedasticity
Answer and working

B. statistic = 7.3; do not reject homoscedasticity

The BPG statistic is half the ESS: 14.6/2 = 7.3 < 7.81, so homoscedasticity is not rejected. Using the full ESS (14.6) would wrongly reject.

Question 5 · hard

A Breusch-Godfrey test for second-order autocorrelation (p = 2) uses n = 60; the auxiliary regression R-squared is 0.14. The 5 percent chi-square critical value with 2 df is 5.99. The test statistic and decision are

  1. 4.20; do not reject no autocorrelation
  2. 8.12; reject no autocorrelation
  3. 0.14; do not reject no autocorrelation
  4. 9.44; reject no autocorrelation
Answer and working

B. 8.12; reject no autocorrelation

BG = (n - p)R^2 = 58 x 0.14 = 8.12 > 5.99, so the null of no autocorrelation up to order 2 is rejected. Dividing by p is not part of the statistic.

Question 6 · hard

Plot area is self-reported with classical measurement error. The true area variance is 80, the measurement error variance 20, and the true slope 0.5. The probability limit of the OLS slope is

  1. 0.40
  2. 0.50
  3. 0.625
  4. 0.10
Answer and working

A. 0.40

With classical error in X, plim b = b x var(X*)/[var(X*) + var(e)] = 0.5 x 80/100 = 0.40: attenuation toward zero. The estimate is inconsistent, not unbiased; it is not inflated. 0.625 inverts the reliability ratio (0.5 x 100/80); 0.10 uses the error share (0.5 x 20/100).

Question 7 · hard

A recursive (triangular) system with uncorrelated errors across equations can be estimated equation by equation with OLS because

  1. no endogenous regressor in an equation is correlated with that equation's error
  2. recursive systems have no endogenous variables, only exogenous ones on the right
  3. OLS is always consistent in simultaneous systems, whatever the error structure
  4. the rank condition fails for recursive systems
Answer and working

A. no endogenous regressor in an equation is correlated with that equation's error

In a recursive system each equation contains only endogenous variables determined in earlier equations; with uncorrelated errors those variables are independent of the current equation's error. That removes simultaneity bias.

Question 8 · medium

Johansen's procedure is applied to wholesale prices from four regional markets, all I(1). The maximum possible number of cointegrating vectors is

  1. 3
  2. 4
  3. 1
  4. 2
Answer and working

A. 3

With n I(1) variables there can be at most n - 1 cointegrating vectors; with n vectors all series would be stationary. Johansen's trace and maximum-eigenvalue statistics test how many there are.

Question 9 · medium

The Granger representation theorem states that

  1. cointegrated variables can always be represented by an error-correction model
  2. any two I(1) series can be written as an ECM, whether or not they are cointegrated
  3. a series that Granger-causes another must be cointegrated with it in the long run
  4. cointegrated series have no short-run dynamics
Answer and working

A. cointegrated variables can always be represented by an error-correction model

Cointegration and error correction are two sides of one fact: a stable long-run tie implies that at least one variable responds to past disequilibrium.

Question 10 · medium

A logit model of hybrid seed adoption gives Z = -2.0 + 0.5 x years of schooling. For a farmer with 6 years of schooling the predicted adoption probability is

  1. 0.731
  2. 1.000
  3. 0.269
  4. 0.841
Answer and working

A. 0.731

Z = -2.0 + 0.5 x 6 = 1.0. P = 1/(1 + e^-Z) = 1/(1 + e^-1) = 0.731. 0.269 is the probability of not adopting; 1.000 is the index itself; 0.841 is the standard normal CDF, which would apply to a probit index.

A study plan for Econometrics in UGC-NET

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 (47% of the 88 questions): Linear probability, logit and probit models (9, 10%); Simultaneous equations and identification (9, 10%); Autocorrelation (8, 9%); Stationarity and unit roots (8, 9%); ARIMA models (7, 8%).

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

  2. Then the middle band: Specification error (7, 8%); Cointegration and error-correction models (6, 7%); Maximum likelihood (6, 7%); Heteroscedasticity (5, 6%); Hypothesis testing in regression (5, 6%).

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

  3. Then the lighter units: Multicollinearity (5, 6%); Distributed lag models (4, 5%); Dummy variables (4, 5%); BLUE properties and the Gauss-Markov theorem (3, 3%); Classical linear regression model and OLS (2, 2%).

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

  4. Leave time for the hard ones: Heteroscedasticity (3 hard of 5), Hypothesis testing in regression (3 hard of 5), Classical linear regression model and OLS (1 hard of 2).

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

  5. Bank the easy marks: BLUE properties and the Gauss-Markov theorem (1 easy of 3); ARIMA models (2 easy of 7).

    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 — UGC-NET Econometrics

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.