Effects of a price limit change on market stability at the intraday horizon in the Korean stock market

ABSTRACT

This article investigates the effects of a price limit change on the volatility of the Korean stock market’s (KRX) intraday stock price process. Based on the most recent transaction data from the KRX, which experienced a change in the price limit on 15 June 2015, we examine the change in realized variance after the price limit change to investigate the overall effects of the change on the intraday market volatility. We then analyse the effects in more detail by applying the discrete Fourier transform to the data set. We find evidence that the market becomes more volatile in the intraday horizon because of the increase in the amplitudes of the low-frequency components of the price processes after the price limit change. Therefore, liquidity providers are in a worse situation than they were prior to the change.

KEYWORDS:

• Price limit
• market stability
• discrete Fourier transform
• intraday price process

JEL CLASSIFICATION:

• C22
• G15
• G18

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ Kim and Park (2010) noted that 23 of 43 important markets use price limit systems.

² Kim and Park (2010) and Deb, Kalev and Marisetty (2010) developed theoretical models that explain the role of price limit systems in the presence of price manipulators.

³ For studies that support proponents of the price limit system, refer to Arak and Cook (1997), Bildik and Elekdag (2004) and Ma, Rao and Sears (1989a, 1989b). For studies that support opponents of the price limit system, refer to Bildik and Gulay (2006), Chen (1998), Fama (1989) and Kim and Rhee (1997).

⁴ For example, Kirilenko et al. (2017) and Menkveld and Yueshen (2016) investigated intraday price processes around and during the Flash Crash.

⁵ Realized variance of Andersen et al. (2001), which is the most popular measure of intraday volatility, assumes that price processes follow geometric Brownian motion.

⁶ According to Malliavin and Mancino (2002), most traditional volatility measures use an algorithm based on a ‘differentiation procedure’ that is highly unstable.

⁷ In our sample period, the regular KRX market is open for 350 min, from 9:00 a.m. to 2:50 p.m. We obtain a 349-dimension price vector array by removing two prices: the opening price at 9:00 a.m. and the price at 2:50 p.m.

⁸ For each w-frequency (w = 1,2…174), the SE is calculated as the SD of $F_{i,w}$ divided by the square root of the number of sample stocks, 200.