A comparison of bid-ask spread proxies: evidence from Borsa Istanbul futures

Abstract

We analyze the performance of five different methods appearing in the market microstructure literature in predicting effectiv e and quoted bid-ask spreads (Roll, LOT Mixed, Effective Tick, High-Low and Closing Percent Quoted Spread proxies). With data from index futures, currency futures and gold futures traded in Borsa Istanbul and taking percent effective and percent quoted spreads obtained from intraday trade and quote data as benchmarks, we calculate and compare the correlations and root mean square errors of the spread measures. Results show that none of the proxies is successful enough in estimating effective or quoted spread although under normal mar ket conditions, Effective Tick appears to perform best.

Liquidity measurement: A comparative review of the literature with a focus on high frequency

“…Starting from the popular ROLL measure (Roll, 1984), which is based upon successive price changes, various models or proxies have been proposed to estimate intraday spread using lowfrequency data (Abdi & Ranaldo, 2017;Corwin & Schultz, 2012;Chung & Zhang, 2014;Fong et al, 2017Fong et al, , 2018Goyenko et al, 2009;Holden, 2009;Lesmond et al, 1999) and several studies have tested their performance and reported that the performance of these proxies are mostly unsatisfactory in terms of their ability to capture spread (Chung & Zhang, 2014;Cobandag Guloglu & Ekinci, 2016;Corwin & Schultz, 2012;Fong et al, 2017;Goyenko et al, 2009;Lesmond, 2005). 10 Moreover, as these proxies mostly aim at measuring daily liquidity, their utility is limited at higher frequencies.…”

Existence of maximal solutions for the financial stochastic Stefan problem of a volatile asset with spread

“…In [7] the German power market liquidity was studied, we also refer to [32] for a statistical analysis of the fluctuations of the average spread where the relation of spread with shares volume and volatility was examined, or to [23] for a stochastic equation model estimating the liquidity risk. In [13], the authors analyzed how transaction costs affect the spreads while in case of zero cost then the market price should act as a Wiener process; see also in [26] for the liquidity risk with respect to the transaction costs and market manipulation under a Brownian motion problem formulation, or in [12,33,21,11], and in [18] for various empirical approaches on spread's forecast. We note that except from the bid-ask spread, there exist several other types of spread like the asset swap spread, the yield spread, the zero volatility spread, the option adjusted spread, the default swap spread, or the bank spreads, see for example in [30,8,9,10,17,22].…”