Advanced Financial Statistics Fall 2021 Problem Set 3
Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit
Fall 2021
Advanced Financial Statistics
Problem Set 3
Individual Assignment, Due Monday 6 December
1. Consider the daily simple returns of the S&P 500 composite index from January 1980 to December 2008. The index returns include dividend distributions. The data file is S&P500WeekDays which has 9 columns. The columns are (year, month, day, SP, M, T, W, H, F), where M, T, W, H, F denotes indicator variables for Monday to Friday, respectively. Use a regression model to study the effects of trading days on the index returns. What is the fitted model? Are the weekday effects significant in the returns at the 5% level? Use the HAC estimator of the covariance matrix to obtain the t-ratio of regression estimates. Does the HAC estimator change the conclusion of weekday effect?
2. The file USMacro_Quarterly contains quarterly data on several macroeconomic series for the United States: the data are described in the file USMacro_Description. Compute Yt = ln(GDPt ), the logarithm of real GDP, and Yt , the quarterly growth rate of GDP. In the problems below, use the sample period 1955:1-2004:4 (where data before 1955 may be used, as necessary, as initial values in regressions).
i. a) Estimate the mean of Yt .
b) Express the mean growth rate in percentage points at an annual rate (Hint: Multiply the sample mean in (a) by 400.)
c) Estimate the standard deviation of Yt . Express your answer in percentage points at an annual rate.
d) Estimate the first four autocorrelations of Yt . What are the units of autocorrelations (quarterly rates of growth, percentage points at an annual rate, or no units at all)?
ii. Estimate an AR(1) model for Yt . What is the estimated AR(1) coefficient? Is the coefficient statistically significantly different from zero? Construct a 95% confidence interval for the population AR(1) coefficient.
a) Estimate an AR(2) model for Yt . Is the AR(2) coefficient statistically significantly different from zero? Is this model preferred to the AR(1) model?
b) Estimate AR(3) and AR(4) models. Using the estimated AR(1)-AR(4) models, use BIC to choose the number of lags in the AR model. How many lags does AIC choose?
ii. Use an augmented Dickey-Fuller statistic to test for a unit autoregressive root in the AR model for Yt . As an alternative, suppose that Yt is stationary around a deterministic trend.
3. There has been much talk recently about the convergence of inflation rates between many of the OECD economies. You want to see if there is evidence of this in North America by checking whether or not Canada’s inflation rate and the United States’ inflation rate are cointegrated.
(a) You begin your numerical analysis by testing for a stochastic trend in the variables, using an Augmented Dickey-Fuller test. The t-statistic for the coefficient of interest is as follows:
Variable with lag of 1 |
InfCan |
∆InfCan |
InfUS |
∆InfUS |
t-statistic |
– 1.93 |
–5.24 |
–2.20 |
–4.31 |
where InfCan is the Canadian inflation rate, and InfUS is the United States inflation rate.
The estimated equation included an intercept. For each case make a decision about the stationarity of the variables based on the critical value of the Augmented Dickey-Fuller test statistic.
(b) Your test for cointegration results in an Engle-Granger Augmented Dickey-Fuller (EG–ADF, see the lecture notes and Stock and Watson, 2007) statistic of (–7.34). Can you reject the null hypothesis of a unit root for the residuals from the cointegrating regression?
(c) Using a working hypothesis that the two inflation rates are cointegrated, describe how you would test whether or not the cointegrating coefficient equals one.
(d) Even if you could not reject the null hypothesis of a unit cointegrating coefficient, would that have been sufficient evidence to establish convergence?
4. In this exercise you will conduct a Monte Carlo experiment that studies spurious regression, a phenomenon where stochastic trends can lead two series to appear related when they are not.
Generate two samples of T=100 i.i.d. standard normal random variables c1,…,c100 and n1,… , n100 . (i) Set Y1 = c1, X1 = n1, and Yt = Yt一1 +ct, Xt = Xt一1 + nt, t=2, …, 100.
(ii) Regress Yt onto a constant and Xt . Compute the OLS estimator, the regression R2 and the t-statistic testing the null hypothesis that the coefficient F1 on Xt is zero.
Use this simulation to answer the following questions.
(a) Run simulation (i) once. Use the t-statistic from (ii) to test the null hypothesis that F1 = 0 using the usual 5% critical value of 1.96. What is the R2 of the regression?
(b) Repeat (a) 1,000 times, saving each R2 and the t-statistic. Construct a histogram of the R2 and the t-statistic. What are the 5%, 50% and 95% percentiles of the distributions of the R2 and the t-statistic? In what fraction of your 1,000 simulated data sets does the t-statistic exceed 1.96 in absolute value?
(c) Repeat (b) for different numbers of observations, for example, T=50, T=200 and T=500. As the sample size increases, does the fraction of times that you reject the null hypothesis approach 5%? Does this fraction seem to approach some other limit as T gets large? What is the limit?
2022-12-02