Elsevier

Journal of Financial Markets

Volume 20, September 2014, Pages 101-128
Journal of Financial Markets

Waiting costs and limit order book liquidity: Evidence from the ex-dividend deadline in Australia

https://doi.org/10.1016/j.finmar.2014.04.001Get rights and content

Highlights

  • Cost of waiting for order execution impacts market quality on the ex-dividend day.

  • Order placement is more aggressive before stocks begin trading ex-dividend.

  • Order submissions are less aggressive on the ex-dividend day.

  • Bid-ask spreads increase substantially on the ex-dividend day.

  • Waiting costs are an important determinant of order choice.

Abstract

In theoretical models of limit order books populated with liquidity traders there is a link between order aggressiveness, spreads, and the cost of waiting for execution. We directly test these models using an experimental setting where waiting time is important for traders, namely the ex-dividend day. Consistent with these models, we find that order placement is more aggressive before stocks begin trading ex-dividend. Stocks with higher expected costs of delaying execution experience larger declines in order aggressiveness from the cum-day to the ex-day. Waiting costs also impact effective bid-ask spreads, which fall on the cum-day before rising on the ex-day.

Introduction

We test the predictions of recent limit order market models developed by Foucault, Kadan, and Kandel (2005) and Roşu (2009) where the cost of waiting to trade is important. Such models are difficult to test empirically because they describe settings where only uninformed participants trade. To overcome these difficulties, we examine how waiting costs impact order choice, order aggressiveness, market depth, and bid-ask spreads around ex-dividend days on the Australian Securities Exchange (ASX).2

Australia׳s imputation tax system plays an important role in the value that investors place on dividends. Corporations are able to provide investors with imputation tax credits attached to dividend payments if the corporation has paid corporate tax in Australia. Investors with a tax liability in Australia are able use the imputation tax credits to reduce their personal tax liability. In effect, the corporate tax paid is a pre-collection of personal tax. As a result, the imputation tax system creates two distinct clienteles. Domestic investors are able to utilize the credits and reduce their personal tax liability. On the other hand, foreign investors without an Australian tax liability will be unable to utilize the credits and will therefore place a relatively lower value on the dividend payment in after-tax terms. To be eligible for the imputation tax credits, investors need to own the security at the close of trading on the cum-dividend day.

This setting has several distinct advantages. First, the ex-dividend day represents a corporate event where traders face a deadline, liquidity-motivated traders are active, and informed traders are not, which is consistent with model settings. Second, the theoretical models describe a setting where traders have subjective differences in valuation, which leads to gains from trade. The differential tax treatment of dividend income for investors provides a distinct institutional setting that drives differences in valuation that are unrelated to private information. Third, theoretical models can be used to describe a market where traders differ in patience, with the relative proportion of patient and impatient traders influencing market outcomes. The ex-dividend day provides a setting where the relative proportions of patient and impatient traders are expected to change in a predictable manner, given the existence of dividend clienteles (Elton and Gruber, 1970, Lakonishok and Vermaelen, 1986, Richardson et al., 1986, Michaely and Vila, 1995, Rantapuska, 2008). With approximately two-thirds of Australian equities held by domestic investors, coupled with an imputation tax system, we anticipate that clienteles will be prevalent in the Australian market and that their trading behavior will change around the ex-day.3 Our analysis of trading and order choice around the ex-dividend day, together with the detailed ASX data for over nearly two decades, provides a unique setting to test theories where liquidity traders face execution risk.

We hypothesize that waiting costs should be higher on the cum-dividend day relative to other days, in light of the impending ex-dividend deadline to trade and the after-tax return that is potentially forgone. For example, Rantapuska (2008) shows that market participants in Finland earn an average overnight return of 2% from the cum-day to the ex-day. Larger waiting costs will induce more competition in the order book from limit order traders, leading to more aggressively priced orders and a narrowing of the bid-ask spread. On the ex-dividend day, waiting costs will be relatively lower in the absence of a trading deadline. With relatively lower waiting costs, the bid-ask spread should be wider as limit orders do not need to be priced as aggressively. We are also able to test our hypotheses in the cross-section of dividend payments, as the level of imputation tax varies across dividend payments from zero to 100%, depending upon the proportion of corporate tax paid on corporation income. If waiting costs are important, we expect the changes in order aggressiveness and spreads between the cum- and ex-days to be more pronounced in stocks whose dividend payments carry a full imputation tax credit. This tax-driven wedge in valuations should encourage liquidity trading and create a higher opportunity cost of not trading for full imputation stocks relative to stocks paying dividends with no imputation. The literature on ex-dividend day valuation and trading also shows that stocks with higher dividend yields and lower transaction costs are likely to attract liquidity traders (Kalay, 1982, Boyd and Jagannathan, 1994), and we anticipate order choice and spread changes to vary with these two characteristics.

Our findings provide broad support for the hypothesis that waiting costs are an important determinant of order choice around a trading deadline and that these costs have a material impact on the aggressiveness of order submissions. The use of market orders is more prevalent on the cum-dividend day, relative to both the ex-day and a 45-day benchmark period prior to the ex-day. After controlling for known factors that impact order placement, it is evident that traders are more aggressive on the cum-dividend day and less aggressive on the ex-dividend day. There is a noticeable asymmetry in the results on the cum-dividend given the increased competition amongst domestic investors seeking to buy before the deadline. For buy orders, the reduction in order aggression between the cum- and ex-day is largest for high yield and full imputation dividend payments, as expected if waiting costs concern traders. We also find that buyers initiate more trades in the cum-dividend period, with the degree of order imbalance positively related to dividend yield and imputation. The depth at the best bid and ask both increase on the cum-dividend day, with ask depth nearly double the depth at the best bid. On the ex-day, best bid and ask depths are similar to the benchmark period. The depths at the top ten bid and ask quotes also increase on the cum-dividend day. However, on the ex-dividend day, the depths decrease significantly at both the top ten bid and ask quotes. The results are consistent with domestic investors buying, and foreign investors selling, on the cum-day prior to the deadline. Both buyers and sellers are trading more aggressively on the cum-day. However, given the higher proportion of domestic investors, we observe that buyers have a tendency to use more market orders, and sellers use more aggressively priced limit orders. This difference in behavior reflects the heightened competition on the buy side, relative to the sell side. On the ex-day, after the deadline has passed, both buyers and sellers are significantly less aggressive in their trading and order submissions.

The theoretical limit order book models of Foucault, Kadan, and Kandel (2005) and Roşu (2009) also imply that bid-ask spreads will be lower when the cost of waiting to trade is high. We find evidence that effective spreads are lower on the cum-dividend day for high yield and full imputation stocks. The hypothesized reduction in waiting costs also has a substantial impact on ex-day spreads. Our results show that the ex-dividend effective half-spread is 10 basis points, or 27%, higher than the average daily effective half-spread in a benchmark period between t−50 to t−6, relative to the ex-dividend day. The effective spread peaks on the ex-dividend day and returns nearly immediately to the average spread level in the days following the ex-day. This finding is robust across dividend yields, imputation levels, and non-event benchmark spread levels. The difference in the effective spread between the cum-day and the benchmark period is positively related to imputation credits after controlling for common determinants of the spread, which indicates that waiting costs differ in the cross-section of dividend payments. Both the price impact and realized spread increase on the ex-day, consistent with tax-motivated liquidity traders shifting their trading to beat the deadline at the close of the cum-dividend day. As uninformed traders move trading to the cum-day that will leave relatively more informed traders and a higher price impact on the ex-day. The increase in the realized spread reflects the decline in waiting costs faced by traders and a resulting increase in trader patience.

This study contributes to the limit order market literature in several ways. We show that differences in valuation by uninformed traders can actually lead to lower spreads in the presence of a trading deadline.4 After the trading deadline has passed and the valuation differential has been removed, spreads actually widen. The existence of this result can be attributed to changes in the aggressiveness of order submissions. Consistent with theoretical models, we also find that the duration between trades decline and the time to execution of limit orders are related to our proxy for waiting costs. In their entirety, our results show that both the costs associated with delayed execution and valuation differentials across traders have a material impact on market quality around a liquidity-driven trading deadline. The current study also has broader implications for predictable changes in trading costs in response to liquidity-motivated trading, such as periods around rights issue ex-dates and benchmark index additions and deletions.

The remainder of the paper is organized as follows. In Section 2, we discuss trading on the ASX and the imputation tax system in Australia. In Section 3, we outline the hypotheses based on the literature regarding dynamic models of the limit order book. In Section 4, we discuss the data used in our analysis. In Section 5, we present the results and in Section 6 we conclude the paper.

Section snippets

Institutional background

The ASX is the major domestic stock exchange in Australia. Prior to October 2, 2006, stocks were traded using the electronic limit order book system called the Stock Exchange Automated Trading System (SEATS). This system commenced operation on October 19, 1987 and fully replaced the trading floor system on September 4, 1990 (Aitken, Brown, and Walter, 1996). SEATS operated until October 2, 2006 when it was replaced by an alternative electronic trading system called the Integrated Trading System

Testable implications of theoretical limit order book models

Our hypotheses are developed from dynamic models of the limit order book presented by Foucault (1999), Foucault, Kadan, and Kandel (2005), and Roşu (2009). These models do not include informed traders, with trading motivated by liquidity needs based on traders with subjective valuations above and below fundamental value. In the ex-dividend setting, the subjective valuations reflect differences between tax rates on capital gains and dividend income, as well as the imputation credit that is

Data and methodology

We obtain intraday trade, quote and order data from the Securities Industry Research Centre of Asia-Pacific (SIRCA) for the period February 19, 1990 to December 31, 2008. This data captures all order submissions, cancellations, and trades that took place on the ASX electronic trading system. Each transaction in the dataset consists of the timestamp to the nearest millisecond, stock ticker, price, volume, bid and ask quotes, bid and ask depth, and trade flags indicating whether the trade was

Order aggressiveness

To address our first hypothesis that liquidity suppliers are more aggressive on the cum-dividend day and less aggressive on the ex-dividend day, we directly examine order aggressiveness to determine whether changes are consistent with our expectations by comparing both days to the average order aggressiveness in a benchmark period from t−50 to t−6. Following Biais, Hillion, and Spatt (1995), Ranaldo (2004), and Comerton-Forde and Tang (2009), we classify orders into six groups of

Conclusion

Theoretical models of limit order books that are populated by liquidity traders indicate that the anticipated cost of the delay between order placement and order execution can impact the resiliency of markets and the choice between market and limit orders by traders in a pure order-driven market. The current study tests the theoretical implications of these models in the ex-dividend period for stocks listed on the ASX between 1990 and 2008. We focus on the ex-day as the presence of tax-induced

References (35)

Cited by (6)

  • CFDs, forwards, futures and the cost-of-carry

    2019, Pacific Basin Finance Journal
    Citation Excerpt :

    As can be seen from Fig. 2, announcements are concentrated around the two to four weeks prior to the expiration of March and September contracts (Panel A). This is expected in Australia as most semi-annual earnings and dividends announcements occur in February and August (e.g. Ainsworth and Lee, 2014). Further, the width of the confidence interval does not seem to increase with the time-to-expiration or fall after the announcement, which suggests that the market anticipates the dividend and dividend uncertainty is not reflected in the variation of the implied dollar dividend.

  • Distilling liquidity costs from limit order books forthcoming in Journal of Banking and Finance

    2018, Journal of Banking and Finance
    Citation Excerpt :

    Hasbrouck and Saar (2013) use order-level data about submissions, cancellations, and executions of orders to measure low-latency activity and to investigate how this variable affects market quality. Ainsworth and Lee (2014) compute market depth from best bid and ask prices to test for the hypothesis that waiting costs impact order choice around a trading deadline. Cao et al. (2009) and Brogaard et al. (2015) propose several summary statistics beyond the best bid and ask to capture features that are amenable for empirical analysis of future price movements at tick level.

  • Trading costs on the Stock Exchange of Thailand

    2015, International Review of Financial Analysis
    Citation Excerpt :

    Furthermore, it is not possible to measure the trading cost for a very large order that uses multiple split orders because the originally desired size is not known. Ainsworth and Lee (2014) find that traders become more aggressive before an ex-dividend deadline because of the higher opportunity cost of not transacting. However, we do not include this event-specific opportunity cost because we are more concerned with the cross-sectional rather than time-series differences.

We thank Sean Anthonisz, Henk Berkman, Carole Comerton-Forde, Tarique Haque, David Johnstone, Gideon Saar (Editor), Tom Smith, Hans Stoll, Avanidhar Subrahmanyam, Kumar Venkataraman (Referee), Terry Walter, Qiaoqiao Zhu and seminar participants at the Australian National University Summer Finance Camp 2010, University of Technology, Sydney, AFAANZ Conference 2010, FMA Asian Conference 2011 and the EFMA 2011 Conference for helpful comments and suggestions.

1

Tel.: +61 2 9514 7765; fax: +61 2 9514 7722.

View full text