Research Colloquium - On the Use of Mixed Sampling in Forecasting Realized Volatility: The MIDAS--MEM
Time
Tuesday, 7. January 2020
15:15 - 16:45
Location
F425
Organizer
Chair of Economics and Econometrics
Speaker:
Giampiero M. Gallo (Rimini Centre for Economic Research)
On the Use of Mixed Sampling in Forecasting Realized Volatility: The MIDAS--MEM
joint work by Alessandra Amendola (University of Salerno), Vincenzo Candila (University of Roma La Sapienza), Fabrizio Cipollini (University of Firenze) and Giampiero M. Gallo (New York University in Florence)
Abstract
When dealing with financial time series, different frequency of observation may reveal relevant information of interest to model market activity. We embed a MIDAS (MI(xed)--DA(ta) Sampling) component in a multiplicative error model (MEM) context (MEM-MIDAS). The proposed specification considers that the conditional expectation of a non-negative process observed daily may accommodate a low frequency component, say monthly. The empirical application is presented on the realized volatility of the S\&P 500, NASDAQ, FTSE 100 and Hang Seng indices. Irrespectively of the series analysed, the MEM-MIDAS statistically outperforms the standard MEM model and other commonly used specifications.