VOLATILITY CONNECTEDNESS AS FEAR GAUGE: SHORT-TERM PREDICTABILITY OF MARKET RETURN AND TRADING ACTIVITY

• Main Author: Arroisi, Abdurrahman
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/36849
• Institution: Institut Teknologi Bandung

Considering the extreme uncertainty faced by investors in the crisis period, this research investigates whether volatility connectedness can be used to predict return and trading activity in the short-term. This research uses transaction data of Indonesian stock market from January 2008 to December 2017. Following the approach of Diebold and Yilmaz (2012), volatility connectedness is constructed using generalized variance decomposition of vector autoregressive model with both high-frequency realized volatility (RV) and low-frequency GJR-GARCH volatility (GGV) as the inputs. By the method of vector autoregression, this research find that GGV-based net volatility connectedness of financial sector can be used to predict future lower market return and lower trading imbalances up to 2 days ahead in the crisis period, but not in the non-crisis period. This result is robust across different portfolios. However, RV-based net connectedness of financial sector has no predictive power both in the crisis and non-crisis. The superiority of GGV-based calculation might be due to leverage effect in the GJR-GARCH. Interestingly, even though with lesser economic significance, RV-based total connectedness (and GGV-based, to the lesser extent) predicts return positively in the 4 or 5 days ahead but not before, in both crisis and non-crisis period. The opposite sign of the return predictability by using net connectedness of the financial sector and total market connectedness indicates a specialty of financial sector in the crisis period. It gives an evidence that financial sector has additional systemic risks that contribute to the future lower market return. Last but not least, we find that Indonesian investors generally perform herding and contrarian investing and that trading activity weakly impact stock return.