/Turn over 821L1: Financial and Time Series Econometrics Assessment Period: May 2021 A2 Duration: 24 hour Take-Away Paper Candidates should attempt ALL questions in section A and ALL question in section B SECTION A 821L1 Financial and Time Series Econometrics 2 ANSWER QUESTIONS 1 AND 2 1. A researcher estimates an earnings function by OLS using a country’sLabour Force Survey data for one year. The dataset contains information on 1,000 individuals. The researcher obtained the following regression estimates (with OLS standard errors reported in parentheses): ln(w)i= 0.612 + 0.05Ei + 0.09Si-0.16Fi + i (0.38) (0.01) (0.03)(0.06) where ln(w) is the natural log of the wage;E isyears of experience; Sis the number of years of schooling, and F is a dummy variable adopting a value of 1 if the individual is female, and 0 if the individual is male. a) Since the data are cross-sectional, what are the potential econometric problems the researcher should worry about when estimating this earnings function and why? [5marks] b) The researcher computes a RESET test and a Breusch Pagan testand obtains the following chi-squaredvalues: RESET= 11.21 ~ ; Breusch Pagan= 0.68~ .Explain how each of these tests is set up. Evaluate the regression model on the basis of these diagnostics. Is there any econometric problem here? If so, what should the next step be for the researcher? [10 marks] c) Precisely interpret the above regression coefficients, commenting in all cases on their signsand statistical significance. [10marks] 2. An investigator is interested in estimating a demand function for electricity. Monthly data on the quantity of electricity consumed (in kilowatt hours) and the price of electricity (in dollars) are available spanning60 months. u 1 1 821L1 Financial and Time Series Econometrics 3 /Turn over a) What are the conditions under which the log of electricity consumption (lec) and the log of the price of electricity (lep) series are stationary? How wouldyou test whether both series contain a unit root? What are the null and the alternative hypotheses in this case? [7 marks] b) The researcherperforms ADF tests with 1 lag and a deterministic time trend for each series (lec and lep) and their first differences (Dlec and Dlep),and obtainsthe following test values: Variables tests ADF t-stat lec -1.32 lep -3.01 Dlec -8.21 Dlep -11.50 Given a 5% critical value of –3.52 for the ADF test for both these cases, which of the series (lec, lep, Dlec and Dlep) are stationary? Determine the appropriate order of integration of lec and lep, Dlec and Dlep. What do you conclude? [8marks] c) The following relationship between the log of electricity consumption and the log ofthe price of electricity is then estimated(with OLS standard errors reported in parentheses)by the researcher: lect = 2.220 - 0.750lept + t (1.077) (0.122) Interpret these estimates. Are the standard errors valid in this case? Explain your answer. [5marks] d) Explain what an econometrician means when they say “lec and lep are cointegrated”. Explain in your own words how an econometrician would test if lec and lep are cointegrated? [5 marks] SECTION B ANSWER QUESTION 3 AND QUESTION 4 u 821L1 Financial and Time Series Econometrics 4 For this section you must formulate, estimate, explain, interpret and report on empirical econometric models. As with any empirical work, you should undertake appropriate diagnostic testing and wherever possible take appropriate remedial action. If you encounter any econometric issues, you must clearly explain the implications these issues may have on your findings. 3. The ‘Exam 1.dta’ dataset contains data from 1956 to1990 onthe logof real GDP (linc), and the logof real consumers’ expenditure (lcons) in the USA. Using the skills and knowledge acquired in class, construct, diagnose, explain and interpret TWO different Error Correction Mechanism models of the relationship between lcons and linc. Which of these models do you prefer and why? [50 marks] 4. The ‘Exam 2.dta’ dataset contains 253 observations of Facebook’s stock price (facebook) and the S&P500 index (SNP500) from 18/04/2017 to 18/04/2018. Recall that news broke on 16/03/2018 of the Cambridge Analytica scandal, where Facebook user data was found to have been harvested and used to influence political outcomes in the US and the UK. Use the constant mean return model to estimate the Cumulative Abnormal Returns for Facebook stock in the 22trading days following news of the scandal breaking. How do these results compare to those identified in class using the market model? Which of these models do you prefer and why? [50 marks] END OF PAPER
欢迎咨询51作业君