Forecasting Methods for Economics — DU B.A. (Hons.) Economics Previous Year Question Papers

Statistical tables are attached for your reference. b) Distinguish between the forecast object and the forecast statement. Explain how the way a forecast is stated can affect decision-making. (5, 5) 2. a) Describe the sales force composite method of forecasting. Briefly explain any two of its advantages and disadvantages. b) A firm has very limited historical data but needs to prepare a short-term sales forecast. Which judgmental forecasting method would you recommend? Justify your answer. (5, 5) The following data relate price (in $’000) to mileage (mpg) for a sample of automobiles: Mileage (X) (mpg) 20 22 25 28 32 a) Estimate the regression model 𝑌𝑌 = 𝑎𝑎 + 𝑏𝑏𝑏𝑏. b) Using the model, forecast the price when 𝑏𝑏 = 27 and construct a 90% prediction interval for the forecast at 𝑏𝑏 = 27. c) Compute the coefficient of determination 𝑅𝑅2and interpret. (5, 5, 5) 4. The following data represent the monthly demand for a product: a) Compute forecasts for periods 4, 5, and 6 using a 3-period moving average. c) Compare the two methods using MSE and comment on which responds faster to changes in demand. 5. The quarterly sales of a product are given below: a) Estimate seasonal indices using the ratio-to-moving-average (CMA) method. b) Deseasonalize the series and identify the trend c) Using the multiplicative decomposition model, forecast the sales for Year 3, Q1. SECTION -III 6. a) Define an AR(1) and MA(1) process. State the condition for invertibility of MA(1). b) Derive the autocorrelation function of an MA(1) process and show that c) Compare the ACF patterns of AR(1) and MA(1) processes and explain how they guide model identification. d) Derive the variance of an MA(1) process and explain how the parameter 𝜃𝜃affects the magnitude of autocorrelations. (5, 5, 5, 5) 7. a) Outline the Engle–Granger two-step procedure for testing cointegration. b) Explain how residual-based unit root testing is used to test for cointegration. c) Discuss limitations of the Engle–Granger approach. d) A long-run relationship between Y and X is estimated as follows: The residuals ûtare tested using an ADF test. The test statistic is -3.4, and the 5% critical value is -3.0. Test whether Y+ and X+ are cointegrated. Clearly state the null hypothesis and the conclusion. (5, 5, 5, 5) 8. a) Define Granger causality and write down the regression equations used to test it. b) State the null and alternative hypotheses for Granger causality. c) Derive the F-statistic used in testing Granger causality and explain its interpretation. d) Consider the following estimated model: It is given that Restricted RSS (excluding X): RSS, = 120 Sample size: n = 30 Number of parameters in unrestricted model: k = 5 Test whether X Granger-causes Yat the 5% level using the F-test. (5, 5, 5, 5) Tis - IlI 620 Statistical tables A: Normal probabilities