A0274
Title: Analyzing intraday variation in price impact: A Bayesian SVAR approach with stochastic volatility estimation
Authors: Makoto Takahashi - Hosei University (Japan) [presenting]
Abstract: The aim is to analyze the intraday variation in the short- and long-term price impact of market orders, limit orders, and cancellations using a structural vector autoregression (SVAR) model. While Bayesian estimation using sign restrictions has been effective in parameter estimation in SVAR models, there are issues with parameter uniqueness. To address this, alternative methods like maximum likelihood estimation and generalized method of moments have been proposed. This study applies a new Bayesian estimation method that considers the stochastic volatility of errors to estimate the model parameters. This method allows the unique identification of the parameters without being affected by the order of the variables by imposing sign conditions on the variables in addition to the heteroskedasticity of the variables. The advantage of this method is that the sign conditions can be easily verified from the posterior distribution of the estimated parameters. This estimation method has not been applied to high-frequency order book data, but the use of a large number of observations allows the model to be estimated every few minutes to tens of minutes and examine the intraday variation. The model simultaneously analyzes the variation in price impact and volatility by modeling and estimating the stochastic variation of both price changes and orders.
