Time-Frequency varying beta estimation - a continuous wavelets approach

Roman Mestre (1) , Michel Terraza (1)

1 MRE - Montpellier Recherche en Economie

The Beta coefficient theorized by the CAPM is estimated by the Market Line. By hypothesis, the Beta is stable over time but empirical studies on it volatility don't confirm this fact. One of them is related to with agent heterogeneity hypothesis. In this paper; we study this hypothesis by continuous wavelets decomposition of the market line components. We use the wavelet Coherence to calculate a time-frequency Beta. We apply this methodology on three French listed stocks (AXA-LVMH-ORANGE) with different OLS beta for the daily period from 2005 to 2015. We show that the coherence and the time-frequency Betas improve our understanding of the equity characteristics and nature according to their time and frequency dynamics. AXA and LVMH have globally an high coherence with the market whereas ORANGE coherence is low (whatever frequencies). These results can affect the time-frequency betas values. By analyzing the betas we see different evolutions and dynamics which can be considered by portfolio managers to optimize their investment horizon. The continuous wavelets is a powerful tool for emphasize the timefrequency instabilities of betas. The hypothesis of heterogeneity of agents have an impact on systematic risk estimations and need to be considered in financial calculations.

Domaines: Pricing [q-fin.PR]

Roman Mestre, Michel Terraza. Time-Frequency varying beta estimation - a continuous wavelets approach. Economics Bulletin, 2018, 38 (4), pp.1796-1810. ⟨hal-03195193⟩