Trong-Nghia Nguyen, Minh-Ngoc Tran, and Robert Kohn, "Recurrent
Conditional Heteroskedasticity," Journal of Applied Econometrics,
Vol. 37, No. 5, 2022, pp. 1031-1054.

The datasets and corresponding Matlab code to clean the data are
zipped in the file RECH-files.zip. Each dataset is stored in an
individual csv file. Description of data files are as follows: 


1. Stock exchange index data

The stock data used in this paper are from the Realized Library
dataset available for download from the website of the Oxford-Man
Institute of Quantitative Finance
(https://realized.oxford-man.ox.ac.uk/). The Realized Library dataset
contains daily non-parametric measures of how volatility financial
assets or indexes were in the past. It consists of daily data of 31
international stock exchange indexes with 18 different realized
measure each. 

For our analysis, we use only daily data for 4 indices: 

- SPX (S&P 500 Index): SPX.csv 
- N225 (Nikkei 225): N225.csv
- RUT (Russel 2000): RUT.csv
- GDAXI (DAX 30): DAX.csv 

Each csv file includes following columns:

- return: The log return series y(t).    
- bv: realized measures of the conditional variance using Bipower
  Variation (5-min)
- rk_parzen: realized measures of the conditional variance using
  Realized Kernel Variance (Non-Flat Parzen)
- rk_th2: realized measures of the conditional variance using Realized
  Kernel Variance (Tukey-Hanning(2))
- rk_twoscale: realized measures of the conditional variance using
  Realized Kernel Variance (Two-Scale/Bartlett)
- medrv: realized measures of the conditional variance using Median
  Realized Variance (5-min)
- rv5: realized measures of the conditional variance using Realized
  Variation (5-min)

In our analysis, we drop rows with missing observations and use the
last 4000 rows of each csv file. The complete (original) data set,
which is downloaded from https://realized.oxford-man.ox.ac.uk/, is
stored in oxfordmanrealizedvolatilityindices.xlsx.

The MATLAB script data_reader.m shows how to extract and clean data
from oxfordmanrealizedvolatilityindices.xlsx to obtain the datasets in
the paper.


2. Currency exchange data

The USD/GBP daily exchange rate data in Section 5.3 are stored in the
file USD_GBP.csv. The data are extracted and cleaned from the original
data stored in realized.library.0.1.csv. The original data are also
downloaded from https://realized.oxford-man.ox.ac.uk/ (Data from
Univariate HEAVY Paper - Shephard and Sheppard). 

The USD_GBP.csv file contains 3 columns:

- GBP_r: log return series
- GBP_rv: realized measures of the conditional variance using Realized
  Variation (5-min)
- GBP_rk: realized measures of the conditional variance using Realized
  Kernel Variance (5-min) 

The MATLAB script data_reader.m shows how to extract and clean data
from realized.library.0.1.csv to obtain the datasets in the paper.


3. Summary of data files:

- data_reader.m : Matlab code to extract and clean datasets used in the
  paper from the original data.
- GDAXI.csv: DAX 30 index data
- N225.csv: Nikkei 225 index data
- RUT.csv: Russel 2000 index data
- SPX.csv: S&P 500 index data
- USD_GBP.csv: USD/GBP exchange data
- oxfordmanrealizedvolatilityindices.xlsx: Original (raw) stock index data
- realized.library.0.1.csv: Original (raw) currency exchange data

The Matlab code showing how to run examples in the paper will be
published to https://github.com/vbayeslab. 


If you have any questions, please contact:

Trong-Nghia Nguyen
nghia.nguyen [AT] sydney.edu.au

Or

Minh-Ngoc Tran
minh-ngoc.tran@ [AT] sydney.edu.au