عملکرد قیمت محدود: شواهد از داده های معاملات و کتاب محدود سفارش
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|9035||2005||22 صفحه PDF||سفارش دهید||11222 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Journal of Empirical Finance, Volume 12, Issue 2, March 2005, Pages 269–290
In recent years, organized stock exchanges with daily price limits adopted wider limits as narrower limits were criticized for jeopardizing market efficiency. This study examines the impact of a wide price limit on price discovery processes, using data from the Kuala Lumpur Stock Exchange. Specifically, examined is the impact of daily price limits on (i) information asymmetry; (ii) arrival rates of informed traders; and (iii) order imbalance. Using both trade-to-trade transaction data and the limit order book, we compile evidence that price limits do not improve information asymmetry, delays the arrival of informed traders, and exacerbates order imbalance. These results suggest that price limits on individual securities do not improve price discovery processes but impose serious costs even when the limit band is as wide as 30%.
In the securities markets, daily price limits represent literal boundaries on where individual security prices are allowed to move, often both upward and downward, and they are typically prespecified by a percentage based on a previous trading session's closing price. Such price limit mechanisms are employed in the U.S. futures markets, but they are also used in many stock exchanges around the world, including Austria, Belgium, France, Italy, Japan, Korea, Malaysia, Mexico, Netherlands, Spain, Switzerland, Taiwan, and Thailand (Roll, 1989, Rhee and Chang, 1993 and Rhee, 2000). Despite their significant presence, however, Harris (1998) contends that we still do not know enough about these market mechanisms to make informed decisions regarding market regulation. Harris (1998) states that appropriate study samples using U.S. data are difficult to obtain. For example, most price limit studies using U.S. futures market data are only able to employ a few contracts (e.g., Chen, 1998 and Ma et al., 1989), which hinders cross-sectional analyses. Consequently, France et al. (1994) state that there are many unanswered questions regarding price limit mechanisms. In our paper, we attempt to provide some much-needed insight into the effects of price limits by investigating transactions data and the limit order book of a stock market that imposes a daily price limit on its individual securities.1 Specifically, we study the price limit system of the Kuala Lumpur Stock Exchange (KLSE) of Malaysia. The primary impetus for studying the KLSE is that we wish to study a market with a wide price limit. The KLSE uses a 30% price limit per trading session. This limit band is much wider than most other exchanges. Recently, several papers have examined the impacts of narrow price limits and have found them to be overly restrictive. For example, Chen (1997) examines Taiwan's previous 7% price limit, Chen (1998) and Park (2000) study the relatively narrow price limits of the U.S. futures markets, Kim and Rhee (1997) investigate the narrow limits of the Tokyo Stock Exchange, and Phylaktis et al. (1999) examine the 4–8% price limit of the Athens Stock Exchange. All of these researchers find potential problems with price limits. However, based on their research, we do not really know whether narrow price limits are bad, or if price limits per se are bad. A policy question that is often raised is “if price limits are to be adopted, then what is the appropriate level?” In fact, conventional wisdom even suggests that wide price limits may be harmless. For example, the Stock Exchange of Thailand recently increased their price limit from 10% to 30%, and the Korea Stock Exchange recently increased their price limit from 6% to 15%. Thus, the impact of wide price limits remains an important policy issue, and a study of wide price limits brings out practical merits for market regulators and for academicians with regulatory policy research interests. The 30% price limit per trading session of the KLSE, therefore, should provide an important first step toward investigating this issue.2 Practical considerations also led us to study the KLSE. Among the few stock markets that we wished to investigate, the KLSE was the only one whose transaction data and limit order book data were available to us. Transactions data and/or limit order books from non-U.S. exchanges are difficult to obtain (and decipher), as reflected by the lack of non-U.S. market microstructure studies.3 Therefore, while this paper contributes to the important expanding market microstructure literature, we also view the fact that our data allow us to study price limit effects at the microstructure level as extremely fortuitous. For example, rather than making inferences on market behavior using only daily data, we will be able to actually observe what happens just prior to, and immediately after, a price-limit-hit. As noted by Lehmann (1989) and Ma et al. (1989), it is difficult to assess price limit effects when relying on only daily data. Finally, as pointed out by Harris (1998), the main reason why price limit studies are “quite scant” is primarily due to the lack of meaningful data. Aside from examining the transactions data and limit order book of a market that employs a wide price limit, our research makes several other very important contributions. First, we examine the impact of price limits on information asymmetry. A popular justification for price limit mechanisms is that they moderate the effects of uncertainty and/or irrationality in the markets by imposing price boundaries. When an “irrational” price reaches its limit, it supposedly provides all traders with time to assess and to recognize the ‘true’ or equilibrium price (for example, the Tokyo Stock Exchange states that their price limits represent “time-out” opportunities). In this context, price limits are supposed to mitigate information asymmetry. However, Amihud and Mendelson (1987, 1991) and Gerety and Mulherin (1992) argue that rational equilibrium prices can only be realized through continuous trading. In addition, informed traders with private information may be unable or unwilling to reveal their information when price limits are hit simply because prices are not allowed to move beyond their limits (Kim and Rhee, 1997 and Kim and Sweeney, 2002). Hence, price limits may not mitigate information asymmetry, but instead, price limits may actually increase information asymmetry. We investigate this issue by examining the degree of information asymmetry before and after a price-limit-hit. If limit-hits do provide time for information dissemination and revelation, then the degree of information asymmetry should be reduced after the limit-hit. Our results, however, indicate that price-limit-hits do not reduce information asymmetry. Our investigation of information asymmetry naturally leads to another, but related, empirical investigation. When the degree of information asymmetry is high, then there is more noise (uninformed) trading (French and Roll, 1986). Uninformed trading leads to price volatility that is unrelated to fundamental value and thus undesirable. Miller (1991) refers to this harmful volatility as episodic volatility, and Harris (1998) refers to it as transitory volatility. During times of uncertainty, if noise trading intensifies, then it may be useful to curb trading. However, the existence of price limits may just as likely exacerbate the noise-trading problem. Kim and Rhee (1997) and Kim and Sweeney (2002) conjecture that information-based trading does not take place during trading sessions when limit-hits occur because rational expectation prices cannot be realized. Instead, informed traders must wait for subsequent trading sessions when price limits have been revised, which mean that information revelation and price discovery is delayed. In the Kim and Rhee (1997) study, they examine price continuation and reversal behavior to see if price limits delay information-based trading. In our investigation, we conduct a much more direct investigation by comparing arrival rates of informed traders before and after a limit-hit. Overall, we find that arrival rates increase after a limit-hit, revealing a delay in information revelation. A third empirical investigation focuses on order imbalance. On one hand, price-limit-hits can be induced by order imbalances, as one-sided supply or demand will drive prices to the limit (Lehmann, 1989). In other words, order imbalances lead to limit-hits. However, it is also possible that the existence of price limits may actually create or exacerbate order imbalances, which, in turn, lead to limit-hits. For example, traders may suboptimally advance their trades in anticipation of a price-limit-hit, which then accelerates the price movement to the limit-hits.4 If traders suspect that trading will cease when prices reach the upper (lower) limit, they will then buy (sell) frantically before the price-limit-hit occurs. This behavior suggests that volume will be one-sided prior to the limit-hit. In other words, the price limits themselves could cause order imbalances. To investigate the impact of price limits on order imbalances, we examine the KLSE order file just prior to and immediately after a limit-hit. We find order imbalances prior to the limit-hit, which is consistent with the view that the impending limit-hit causes order imbalances. For example, for a control group that also experienced a large price change, but without a limit-hit, we do not find a similar order imbalance just prior to its large price change. Further, when we look to the period immediately after limit-hits, we find order imbalance reversals (i.e., an order imbalance exists, but in the other direction). This reversal activity further suggests that traders are attempting to correct for (reverse) their earlier suboptimal trades. If the prehit trades had been not suboptimally executed in anticipation of an impending limit-hit, then an order imbalance during the postlimit-hit period would not have been observed. Overall, this evidence suggests that limit-hits disrupt the liquidity of the markets and cause order imbalances. Finally, we should mention that the past empirical work on price limits has focused primarily on volatility. The recent literature is beginning to converge toward the opinion that price limits do not moderate volatility (for example, see Chen's, 1998 and Park's, 2000 studies on U.S. futures markets, Kim's, 2001 study of the Taiwan Stock Exchange, Kim and Rhee's, 1997 study of the Japanese market, and Phylaktis et al.'s, 1999 study of the Greek market). However, if excessive volatility is an outcome of irrational behavior and if price limits exacerbates this, as the literature suggests, then it may be just as important, if not more important, to focus on exactly why price limits cannot reduce harmful volatility. Prior papers do not explicitly investigate this issue, leaving the relationship between price limits and volatility as somewhat ambiguous. In our paper, we provide some important empirical evidence on the link between price limits and excessive volatility. If price limits do not reduce information asymmetry, but instead delays information trading and contributes to the order imbalance, then this reveals how and why price limits are ineffective in reducing transitory volatility. Therefore, what further differentiates our paper from others is that we address important questions that have been raised regarding price limits but not yet been investigated. The rest of this study is organized as follows. Section 2 discusses the institutional background of the KLSE. Section 3 describes our data and presents summary statistics. Section 4 outlines our empirical design and findings. The last section summarizes the results and presents concluding remarks.
نتیجه گیری انگلیسی
This paper studies the Kuala Lumpur Stock Exchange's 30% price limit system. Prior papers have criticized narrow price limits, and, consequently, these studies can only suggest that narrow price limits, not price limits per se, are bad (which is a rather unsurprising finding). An ‘optimal’ price limit range, if it exists, could occur at wider price limits. By examining the KLSE's wide price limit band, we take an important first step towards addressing this issue. In addition, the KLSE is a useful market to study because its market characteristics (e.g., volatility and turnover activity) are similar to other markets, making our findings potentially applicable to other markets. At the same time, the KLSE employs a call auction trading system, giving us an opportunity to see price limits effects under a somewhat unique market structure. Finally, our study makes use of transactions data and the limit order book. Thus, we are able to contribute to the expanding market microstructure literature, while adding to the price limit literature with more meaningful data. In this paper, we address new research questions regarding price limits. Specifically, we examine the impacts of price limits on information asymmetry, arrival rates of informed traders, and order imbalance. In conducting our study, we first identify a study sample of stocks that actually hit their price limit. By comparing the pre- and posthit periods, we can then identify the impact of the limit-hit. However, we are well aware that any observed differences between the pre- and posthit periods could be associated to (i) the price-limit-hit or (ii) the large price change. Therefore, we create a control sample of stocks that also experience a large price change but did not hit their limit. By looking at pre- to posthit period changes within each stock group and by comparing these changes across the two stock groups, we can better identify the impacts of price limits. Specifically, we find that price limits (1) do not improve information asymmetry, (2) delay the arrival of information, and (3) cause order imbalances prior to and after a limit-hit. In conclusion, our results reveal that even in a market with a wide price limit band, price limits do not improve market efficiency but impose serious costs. Recently, the Stock Exchange of Thailand expanded their price limits from 10% to 30%, but the Taiwan Stock Exchange narrowed their price limits from 7% to 3.5%. However, based on prior research and our own empirical study, trying to identify the optimal price limit level may be a futile task. Instead, policy makers may wish to consider eliminating these price limit mechanisms.