MSIN0209: Finance Research Project
Finance Research Project
MSIN0209: Finance Research Project – Research proposal example
Volatility in the FTSE 100 in bull markets vs bear markets: Applications for investors
1. Introduction
The state of the market, a bull or a bear market, has an important role in financial
decision-making. The terms bull and bear mean extended intervals of time over which prices
have broadly increased or broadly decreased respectively (Chauvet and Potter, 2000).
To identify and forecast bull and bear markets, semi-parametric and parametric methods
are often employed. Pagan and Sossounov (2003) developed a semi-parametric method that
uses a range of rules to determine bear and bull markets in equity prices. Hamilton (1988,
1989, 1990) developed a parametric method, the Markov regime-switching model, that is
used to forecast prices based on switching regimes that take the volatility of returns into
account. An advantage of the semi-parametric methods is that they use transparent rules to
determine the peaks and throughs. Nevertheless, the rules require some subjective settings,
while also predictor variables that are relevant for forecasting the market prices are not used.
The regime switching models specify one data-generating process, allowing a complete
density of returns at each point in time. However, the method is vulnerable to
misspecification, because of the need to specify the data-generating process.
Even though both these methods have been used in various market (such as the S&P500),
no study has assessed the efficacy of these two methods in the FTSE100 index. Motivated by
the unique features of the FTSE100 (e.g. time-varying volatility), the aim of this study is two-
fold: First, to compare between semi-parametric and parametric forecasting methods for the
UK market proxied by high-frequency FTSE100 data over the period 1980 – 2019. To enhance
the robustness and appropriateness of the models several macroeconomic and financial
variables are used to predict switches between bull and bear markets. Second, the study
attempts to identify which model provides the best measures of conditional time-varying
variance of the FTSE100 in bull and bear markets.
2. Theoretical Background and Hypotheses development
The market state is monitored because of its relation to (a) the credit supply (Rigobon and
Sack, 2003, Bohl et al., 2007), (b) time-variation in risk premia (Gordon and St-Amour, 2000,
Ang et al. 2006) and (c) forecast macroeconomic variables and predict business cycle
(Marcellino, 2006). Therefore, I hypothesize that:
H1: Parametric models are more suitable to forecast bear and bull market compared to semi-
parametric models in high frequency and extremely volatile data, such as the FTSE100.
H2: GARCH (p,q) fits better the FTSE100 compared to EGARCH(p,q) and TGARCH(p,q) and
H3: EGARCH (p,q) fits better the FTSE100 compared to TGARCH(p,q).
3. Possible data sources
I will use the FTSE 100 data of the period 1980 to 2019, found on Yahoo Finance.
4. Empirical Analysis
First, I will identify which periods from 1980 – 2019 were bullish or bearish. As a semi-
parametric method, this study uses the Pagan and Sossounov (2003) method to determine the
past states of the market. Hamilton’s (1990) method is used as a parametric method.
Additionally, due to the extremely volatile nature of the FTSE 100 I will attempt to identify the
best model to measure conditional variance of the FTSE100. The approach is suitable
when a series exhibits volatility clustering and serial correlation, suggesting that past variances
might be predictive of the current variance. Thus, I will use the GARCH(p,q), EGARCH(p,q) and
the TGARCH(p,q) to identify the best model to capture volatility during bear and bull periods.
4. Applications
Investors and economists need robust information about the future state of the market. The
results of this project could support investment decisions during bear and bull periods.