Modelling stock returns in an OECD country
The objective of this assignment is to estimate some alternative models of stock returns using data on one of the OECD countries. The assessment will be based on ability to analyse econometric results and carry out appropriate hypothesis testing. All the model building can be carried out using the software package E-Views.
The Data Set
The data set consists of monthly share price indices for 30 OECD countries over the period January 1960 to February 2002. (For some countries the available data period may be shorter).
1 AUSSP AUSTRALIA: SHARE PRICES S&P
2 AUTSP AUSTRIA: Share prices: VSE WBI index
3 BELSP BELGIUM: Share prices: All Shares index
4 CANSP CANADA: Share prices: TSE 300 composite index
5 CHESP SWITZERLAND: Share prices: UBS 100 index
6 CHNSP CHINA: Shanghai Composite Index (19/12/1990)
7 CZESP CZECH REPUBLIC: Share prices: PX-50 index
8 DEUSP GERMANY: Share prices: CDAX index
9 DNKSP DENMARK: Share prices: KAX CSE All Shares index
10 EMUSP Euro area: Dow Jones EURO STOXX Broad
11 ESPSP SPAIN: Share prices: MSE General Index
12 FINSP FINLAND: Share prices: HEX All Share index
13 FRASP FRANCE: Share prices: Paris Stock Exchange SBF 250 index
14 GBRSP UNITED KINGDOM: Share prices: FTSE-A Non-Financial
15 GRCSP GREECE: Share prices: ASE Composite index
16 HUNSP HUNGARY: Share prices: BUX Share price index
17 IRLSP IRELAND: Share prices: ISEC index overall
18 ISLSP ICELAND: Share prices: ISE Share Price index
19 ITASP ITALY: Share prices: ISE MIB Storico
20 JPNSP JAPAN: Share prices: TSE Topix index
21 KORSP KOREA: Share prices: KSE KOSPI index
22 MEXSP MEXICO: Share prices: MSE Share price index
23 NLDSP NETHERLANDS: Share prices: All shares index
24 NORSP NORWAY: Share prices: index OSE Total TOTX
25 NZLSP NEW ZEALAND: Share prices: NZSE Capital Index 26 POLSP POLAND: Share prices: WIG all shares index
27 PRTSP PORTUGAL: Share prices: BVL general share price index
28 SWESP SWEDEN: Share prices: AFGX Index
29 TURSP TURKEY: Share prices: ISE National - 100
30 USASP UNITED STATES: Share prices: NYSE Common Stocks
The data set is available in the file oecd.wf1 on the ULearn page for this module.
The allocated country to analyse is JAPAN.
(a) Create a log return series from the stock price index (the log return is the change in the logarithm of the stock price). Estimate an appropriate pure AR(p) model and an appropriate pure MA(q) model and compare these to a mixed ARMA(1,1) model. Carefully explain how your models were selected. (40%)
(b) Using your preferred model from (a), test for the presence of any ARCH effects in the residuals and estimate an appropriate GARCH model. (30%)
(c) By selecting an appropriate model in the GARCH family, test for whether there is any evidence of:
(i) a link between risk (conditional volatility) and the log return (15%)
(ii) an asymmetric response to positive and negative volatility shocks. (15%)
Credit will be given for full explanation and discussion of methods and results. Numeric answers alone will only score minimum marks. The assignment should be written up in not more than 2000 words. Collusion will be severely penalised.