Non-trading, market making, and estimates of stock price volatility

Abstract

We examine the effects of market making and intermittent trading on estimates of stock price volatility. When observed price changes are correctly tied to a stock's true price dynamics, it is found that nontrading per se causes a loss of efficiency but no bias in traditional volatility estimates. Nontrading induces substantial inefficiency in the extreme value estimator of volatility which it biases downward. Market making's effects add to the nontrading-induced inefficiency in the traditional estimator, while information trading imparts a downward bias, and liquidity trading a potentially removable upward bias, In that estimator.Address proofs and reprint order form to

Forecasting performance of extreme‐value volatility estimators

“…Therefore, the EV estimators are more appropriate even for the daily volatility estimation. The early literature generally confirms a negative bias in EV estimators, as compared to historical volatility (for instance, Garman & Klass, 1980;Marsh & Rosenfeld, 1986;Wiggins, 1991Wiggins, , 1992. We find them positively biased compared to the historical volatility for daily estimations.…”

“…Early empirical studies on these estimators (Wiggins, 1991(Wiggins, , 1992 confirmed their relative efficiency in estimation but raised doubts about their bias, particularly for less liquid assets. Marsh and Rosenfeld (1986) also reported a bias in these estimators due to discrete trading. However, these studies used historical volatility as the benchmark.…”

Estimating the volatility of S&P 500 futures prices using the extreme‐value method

“…The Parkinson (1980) estimator in particular contains little empirical bias relative to close-to-close volatility, suggesting the discrete-time and discrete-price trading problems discussed in Garman and Klass (1980) and Marsh and Rosenfeld (1986) are not particularly serious for an actively traded instrument with small price increments.…”

Analysis of Some Volatility Estimators: An Application to Historical Data from the Nigerian Stock Exchange Market

“…In spite of these appeals, the strict assumptions of log-normal asset returns distribution and continuous trading have been a big obstacle for attracting enough attention. For instance, Marsh and Rosenfeld [18] and Wiggins [27][28] show that the analyses using range volatility estimators succeed in enhancing efficiency but fail to reduce biasness. The main cause of this finding is the low liquidity of the assets under their studies which lead to the violation of continuous trading assumption.…”