CS代写 Microsoft Word – Tutorial 8 T2 2021.docx

Microsoft Word – Tutorial 8 T2 2021.docx

Copyright By PowCoder代写 加微信 powcoder

Financial Econometrics T2 2021
Tutorial 8
1. (Miscellaneous questions)

(a) Why do we care about the conditional variance matrix of a vector of returns? What

are required for a matrix to be a “variance matrix”?

(b) What is the daily realised variance of an asset return? What does it estimate?

2. (Realised volatility, data source: http://realized.oxford-man.ox.ac.uk/data)

This exercise is based on on the data in sp500rv-tut11.xls. The time series are related

to the S&P500 index. There are 3 columns in the file: Date, RV (daily realised variance of

the index, 5-minute intervals), Close (daily close price of the index), spanning from 2000-01-

03 to 2012-10-05 with a total of 3181 observations. The goal of this exercise is to examine

the empirical characteristics of the realised variance.

(a) Compute the daily return R

Compute the daily log realised variance LRV

Compute the daily realised standard deviation RSD by typing

and the standardised return SR=R/RSD

Make the time series plots of R, LRV and SR and comment. Find the descriptive

statistics and histograms of R, LRV and SR and comment. Here, “comment” means

“describe the features of the associated objects”.

(b) Find the correlograms of R and LRV and comment.

(c) Is there a unit-root in LRV? If you are to fit an ARMA(𝑝, 𝑞) model to LRV, what

would be the 𝑝 and 𝑞? Estimate your ARMA model, check the residual for autocorrelation

and comment. Also is the residual normally distributed?

(d) Define the residual from the model in part (c) as E, and the fitted (or predicted) value

as F = LRV−E. The conditional variance of R can be approximated by CV = A × exp(F)

where A is the sample mean of exp(E). Then, the conditional standard deviation CSD is

computed as the square root of CV. Make a scatter plot of RSD agains CSD and comment.

(e) Estimate an AR(1)-EGARCH(2,1) model for the return R, which only relies on the

return series. Define the conditional standard deviation from the EGARCH(2,1) as EGSD.

Make time series plots EGSD and CSD, as well as a scatter plot of EGSD against CSD.

Compute the cross correlation of EGSD and CSD and comment.

(f) Re-estimate the AR(1)-EGARCH(2,1) model for R, including LRV(-1) in the

variance equation, ie,

ln(𝜎!”) = 𝛼# + 𝛼$|𝑣!%$| + 𝛾𝑣!%$ + 𝛼”|𝑣!%”| + 𝛽$ ln(𝜎!%$” ) + 𝜓LRV!%$.

Compare the estimation results to those in part (e) and comment.

程序代写 CS代考 加微信: powcoder QQ: 1823890830 Email: powcoder@163.com