Document Sample

Market Quality Surrounding a Tick Size Increase: Evidence from the Sydney Futures Exchange ANTHONY FLINT, DIONIGI GERACE and ANDREW LEPONE* Finance Discipline, Faculty of Economics and Business, University of Sydney, Sydney, 2006, Australia * This research was funded by the Sydney Futures Exchange under Corporations Regulation 7.5.88(2). Market Quality Surrounding a Tick Size Increase: Evidence from the Sydney Futures Exchange Abstract Numerous studies have examined the relationship between tick size reductions and liquidity under different market structures. This study is the first to examine the impact of a tick size increase on market quality in a futures market setting. It uses the natural experiment provided by the tick size increase for 3-Year Commonwealth Treasury Bond Futures on the Sydney Futures Exchange in 2009. Results show an increase in bid-ask spreads following the change, in addition to a significant increase in quoted depth at the best quotes and throughout the limit order book. Bid-ask spreads per minimum tick declined, as quoting around the minimum tick is tighter. Execution costs are shown to be significantly higher after the change. We conclude that an increase in the tick size has a negative impact on liquidity. 1. Introduction Financial exchanges worldwide have lowered the minimum price increment to facilitate greater levels of liquidity to market participants. Reductions in the tick size do not always generate improvements in market quality if it leads to a pricing grid that is too fine. A change in tick size has important implications for the bid-ask spread, the level of depth, trading and order submission strategies, the aggregate of which may lead to an increase or decrease in overall transaction costs. Recently, a number of exchanges have begun to increase the minimum tick for certain interest-rate futures.1 The stated reason for this change is that the higher tick size is expected to encourage quoting on the part of traders as placing orders is more profitable. This paper examines the impact of a tick size increase on market quality under a futures market setting. The decision by major exchanges to reduce the minimum tick has lead to considerable interest on the part of academic researchers on the relationship between tick size and transaction costs. Harris (1994) argued that a finer pricing grid would lower the quoted spread as the tick size constitutes the lower bound of the spread, while quoted depth would fall as the marginal profitability of supplying liquidity is reduced. The empirical evidence on tick size reductions support the predictions of Harris (1994), finding both lower quoted spreads and lower quoted depth, though the literature conflicts on which has the larger impact on liquidity. Examining the reduction in tick size for stocks priced below five dollars on the American Stock Exchange, Ahn et al., (1996) document a 19% 1 The Sydney Futures Exchange increased the size of the minimum tick on the 3-Year Commonwealth Treasury Bond Futures from 0.5 to 1 basis point. The Eurex increased the Euro Bobl Futures Bond from 0.5 percent to 1 percent on the 30th August, 2009. The CME Group increased the tick size of the 30-Year US Treasury Bond from 0.5 to 1 basis point on the 30 th August, 2009. decline in spreads for the most actively traded stocks, while market depth is unaffected. In another study, Ahn et al., (1998) find the reduction in tick size on the Toronto Stock Exchange (TSE) to five cents for stocks cross-listed on NYSE/AMEX decreases quoted spreads by 27% and share volume by 19.9%, respectively. Supporting Ahn et al., (1998), Bacidore (1997) report a decline in both spreads and depth on the TSE with no adverse affect to large investors, concluding market quality remained unchanged after the tick decrease. In an analysis of the 1997 reduction in the minimum price increment from an eighth to a sixteenth on the NYSE, Ricker (1998) reports a decline in volume-weighted bid-ask spreads of 26% and a decline in quoted depth of 45%. The author concludes the reduction lead to an improvement in liquidity, particularly for low-priced shares. Studying the same event, Goldstein and Kavajecz (2000) find reduced transaction costs for small market orders but higher trading costs for larger market orders, concluding the reduction in tick size did not improve liquidity for all market participants. The results were sensitive to how actively the security was traded, with the benefit to small orders sharply reduced for infrequently traded stocks. Johnson and Lipson (2001) show higher transaction costs for institutional orders after the decrease in tick size on the NYSE, with orders above 100,000 shares facing increased transaction costs of a third more after the tick size reduction. Bessembinder (2003) reports reduced spreads and depth on the NYSE and NASDAQ after the move to decimalisation in 2001, concluding that liquidity increased. In contrast to Jones and Lipson (2001), Chakravarty et al., (2005) associate the move to decimalisation with decreased transaction costs to institutional investors and a significant decline in trading costs overall. Bourghelle and Declerck (2004) examine a change in the pricing grid on the Paris Bourse which raised the tick size for certain stocks and lowered it for others. The authors reveal the reduction (increase) in the tick size is associated with a decrease (increase) in quoted depth, while investors use more (less) hidden orders after the decrease (increase) in tick size. The results show no change in relative quoted and effective spreads under both an increase and decrease in tick size, suggesting a convex relationship between the tick size and bid-ask spread. They conclude that reducing the tick size is not always optimal as a coarse pricing grid may not lead to excessively large spreads, increases quoted depth and encourages liquidity providers to expose their trading interest. Investigating the move to decimalisation for UK Long Gilt Futures on LIFFE, ap Gwilym et al., (2005) report increased quoted spreads and a fall in the mean trade size after the change. Alampieski and Lepone (2008) examine the reduction in the minimum price increment of the 3-Year Commonwealth Treasury Bond Futures (3-Year bond futures) on the SFE. Results indicate a significant reduction in both bid-ask spreads and quoted depth. The authors estimate price impacts of orders which are found to decrease after the change, indicating an improvement in liquidity as the reduction in bid-ask spreads offset the reduced levels of quoted depth. This paper contributes to the literature by investigating the impact of the increase in minimum tick of the 3-Year futures bond on the Sydney Futures Exchange (SFE) on market quality. Contrary to the proponents of tick size reductions, a coarser pricing grid may have a positive impact on liquidity. In the theoretical model of Cordella and Foucault (1999), the price increment which minimizes the cost of immediacy is not zero. If the current tick size is too fine, an increase in the minimum tick will increase the propensity of investors to post at the competitive spread. Second, even considering that a higher minimum tick increases the cost of immediacy, this may be offset by significant growth in limit orders leading to an overall improvement in market quality. We examine market quality indicators such as the bid-ask spread, quoted depth, traded volume and volatility before and after the change in tick size in the 3-Year bond futures on the SFE. As changes in liquidity can result from changes in market conditions as opposed to the increase in minimum tick, the substitute 10-Year Commonwealth Treasury Bond Futures (10-Year bond futures) is used as a control. An issue with prior studies showing declines in both spreads and depth is determining which has a greater impact on liquidity. We estimate the price impact of trades before and after the change in tick size to examine the overall change in liquidity. Results suggest that bid-ask spreads and quoted depth increase significantly after the change, while there is a reduction in bid- ask spreads per minimum tick. Execution costs are found to increase significantly. These results hold in overnight trading and are robust to market-wide factors. This suggests that a higher tick size reduces liquidity under a futures market setting. The remainder of this paper is structured as follows. Section 2 describes the data and empirical results and section 3 concludes. A brief overview of the institutional details of the Sydney Futures Exchange are provided in the Appendix. 2. Data and Empirical Results The data used in this study is provided by Securities Industry Research Centre of Asia Pacific (SIRCA) and contains a record describing each transaction, including the contract code, date, time and the price and volume of each trade. The data also provides the prices and volumes of prevailing bid and ask quotes throughout the limit order-book time stamped to the nearest second. On the 11th of May 2009 the SFE increased the minimum tick size from 0.5 to 1 basis-point for the 3-Year bond futures contract.2 To examine the impact of the increase in minimum tick on market quality, we examine two subsamples 3 months before and 3 months after the change. The pre-period is 10 February, 2009 to 10 May, 2009 and the post-period is the 13 May, 2009 to 13 August, 2009. As the change in tick size occurred at 17:00 hours on the 11th of May, 2009, both the 11th and the 12th of May are excluded from the sample. In line with Frino and McKenzie (2002) who find abnormal levels of trading near expiry, we exclude the 5 days prior to expiration. In line with Bortoli, Frino, Jarnecic and Johnstone (2006), analysis is restricted to the nearest to expiry contract only. 2 The change in tick size on the 11th of May 2009 coincided with a reduction in the visibility of the order book in the 3-Year bond futures from 5 to 3 price levels. Changes in liquidity before and after the increase in minimum tick may reflect changes in market conditions as opposed to the change in tick size. To determine whether this is the case, the 10-Year Commonwealth Treasury Bond Futures (10-Year bond futures) is used as a control contract. The two futures contracts are regarded as substitutes as they trade on the same platform during the same hours, with underlying assets being risk-free Australian Government bonds. The minimum tick size of 0.5 basis points and the level of transparency of 5 price levels on each side of the order-book for the 10-Year bond futures remained constant over the sample period. The variables used to assess changes in market quality before and after the change to full- basis point trading in the 3-Year bond futures are the bid-ask spread, quoted depth, traded volume and volatility. The bid-ask spread is calculated using two measures. The first is the absolute bid-ask spread in points, measured as the ask-price minus the bid-price. Following Frino, Lepone and Wearin (2008), the second measure employed divides the absolute bid-ask spread by the minimum tick. The bid-ask spread is sampled over 5- minute (15-minute) intervals for the day (night) and then averaged over each trading day. Lee et al (1993) establish that an examination of liquidity must involve an analysis of both spreads and depth, with Harris (1994) arguing that changes in liquidity can only be determined by assessing changes in depth throughout the limit order book. Goldstein and Kavajecz (2000) note that an analysis of depth at the best prices omits valuable information as to whether the change in tick size results in a sufficient change to cumulative depth to change the transaction costs of large orders. Consequently, quoted depth is examined using two measures; best depth and total depth. Best depth is defined as the combined volume of shares available at both the best bid price and best ask price at the end of each interval. Total depth is the sum of the volume of shares at each bid and ask price throughout the visible limit-order book at the end of each interval. Similar to bid-ask spreads, best and total depth are sampled over 5-minute (15-minute) intervals for the day (night) and then averaged over each trading day. Traded volume is calculated as the total number of shares traded during the day. Volatility is measured as the natural logarithm of the highest traded price divided by the lowest traded price for each day. 2.1 Univariate Results Table 1 provides descriptive statistics of the market quality indicators for the 3-Year and 10-Year bond futures before and after the increase in minimum tick size. Prior literature including Ahn et al., (1996, 1998), Goldstein and Kavajecz (2000) and Alampieski and Lepone (2008) indicate reductions in the minimum tick lead to lower bid-ask spreads. Supporting this empirical evidence, bid-ask spreads for the 3-year Treasury bond are significantly wider after the increase in minimum tick. Average bid-ask spreads increase from 0.0053 basis points in the pre-period to 0.0102 points in the post-period for trading during the day, and from 0.0089 basis points to 0.0119 basis points for night trading. Bid- ask spreads as a function of the minimum tick are tighter after the change however, decreasing from 1.055 ticks in the pre-period to 1.015 ticks in the post-period for day trading, a statistically significant decline. A similar result occurs for night trading. The reduction in spreads per minimum tick may reflect an increasing proportion of traders quoting at the competitive spread as it is more profitable to trade given a larger tick size.3 Results for the 10-Year bond indicate a significant decline in both measures of the bid- ask spread for day and night trading after the change. This suggests the reduction in spreads per minimum tick for the 3-Year futures contract could be driven by market-wide improvements in liquidity. The increase in the absolute bid-ask spread however is confined to the 3-Year contract. Harris (1994) predicted a reduction in the tick size would decrease quoted depth as liquidity provision is less profitable and more risky. Supporting this conjecture, depth at the best prevailing quotes and total depth throughout the visible order book exhibits significant growth after the change for the 3-Year bond futures. Depth at the best prevailing quotes increases by a significant 959 contracts during day trading and a significant 255 contracts during night trading. Total depth in the order book follows a similar trend, increasing from 1833 contracts (1008 contracts) during the day (night) to 4686 contracts (1610 contracts). This result is consistent with Alampieski and Lepone (2006) who report a decline in quoted depth on the SFE after the reduction in the minimum tick in 2006. While this indicates that the change in tick size has improved the level of quoted depth, depth for the 10-Year bond futures contract shows a similar increase after the move to full basis-point trading. There is a statistically significant increase of 37 contracts at the best prevailing quotes and 210 contracts for total depth (best depth increases by 10 3 As a test of this assertion, we calculate the proportion of minimum tick spreads to the total number of minimum tick spreads during each trading day and night over the two sample periods. A paired t-test shows the proportion of minimum tick spreads quoted increases significantly after the tick size change. contracts and total depth increases 44 contracts in overnight trading). These results suggest the improvement in depth levels for the 3-Year Bond futures could also be driven by market-wide factors, not the change in tick size. The level of average daily volume for the 3-Year bond futures displays a significant improvement after the tick increase, growing from 39,665 contracts in the pre-period to 51,461 contracts in the post period. A similar significant increase of 5,381 contracts occurs for the overnight session. There is a decrease in traded volume for 10-Year bond futures, though this movement is statistically insignificant. In contrast, overnight volume for the 10-Year bond futures contract shows an increase of 981 contracts, which is significant at the 1 percent level. There is no significant change in volatility for both futures bonds. To further ensure the change in liquidity in the 3-Year bond contract results from the increase in the minimum tick and not the impact of seasonality in trading, the post period is compared to the period 13 May, 2008 to 13 August, 2008. Descriptive statistics presented in Table 2 indicate that the change in liquidity is driven by the increase in tick size and not market-wide impacts. Supporting earlier results, there is a significant increase of 0.005 basis points in the bid-ask spread for the 3-Year bond, while spreads as a function of the minimum tick declined by a significant 0.0215 ticks. For the 10-Year bond, both raw spreads and spreads per minimum tick declined for day trading. There was a statistically significant increase in bid-ask spreads for night trading, though less than the 3-Year bond (0.0008 basis point increase for the 10-Year bond as opposed to the 0.004 basis point increase for the 3-Year bond). Quoted depth at both the best quotes and throughout the limit order book increased for the 3-Year bond, in contrast to the 10-Year bond where depth levels declined. Trading volume is significantly higher for the 3-Year bond futures but significantly lower for the 10-Year bond, while volatility is significantly higher across both contracts. The greater levels of trading volume may have been caused by the tick increase, as the higher profit potential from wider bid-ask spreads attracts traders to the market. Controlling for seasonal trading patterns, the increase in the bid-ask spread and depth levels are confined to the 3-Year contract. 2.2 Regression results As documented by Chordia, Roll and Subrahmanyam (2000), changes in market quality indicators such as bid-ask spreads and quoted depth covary with changes in market-wide liquidity factors. To control for broad market movements that captures market-wide factors affecting bid-ask spreads and quoted depth, the following regressions are estimated BAS3 = α0 + α1Change + α2BAS10 + ε (1) Ln(BestDepth3) = α0 + α1Change + α2Ln(BestDepth10) + ε (2) Ln(TotalDepth3) = α0 + α1Change + α2Ln(TotalDepth10) + ε (3) where Change is a dummy variable assigned the value of 1 if the observation is taken from the post-event sample, or 0 otherwise, BAS3 is the bid-ask spread per minimum tick, Ln(BestDepth3) is the logarithm of the aggregate order volume at the best bid and best ask price and Ln(TotalDepth3) is the logarithm of the aggregate order volume throughout the limit-order book for the 3-Year bond futures. The variables BAS10, Ln(BestDepth10) and Ln(TotalDepth10) represent bid-ask spreads, the logarithm of best depth and the logarithm of total depth for the 10-Year bond contract, respectively. In equation (1) the bid-ask spread per minimum tick as opposed to the absolute bid-ask spread is used as the dependent variable. One of the major reasons for the current trend on exchanges to increase the tick size is to encourage quoting on the part of liquidity suppliers. It is more informative therefore to examine the spread per minimum tick as it indicates whether an increasing proportion of traders are quoting at the competitive spread. These three regressions are re-estimated using the 10-Year bond futures as dependent variables. Chordia, Roll and Subrahmanyam (2000) also show that liquidity measures such as the bid-ask spread and quoted depth are dependent on factors specific to the particular financial instrument in addition to market-wide liquidity factors. Harris (1994) argues that two important determinants of the bid-ask spread and quoted depth are traded volume and price volatility. To control for both market-wide and security specific factors on the bid-ask spread and quoted depth, we follow Frino, Gerace and Lepone (2008) and estimate the following equations BAS3= α0 + α1Change + α2 Ln(Volume10) + α3Volatility10 + α4 Ln(Volume3) + α5Volatility3 + ε (1) Ln(BestDepth3) = α0 + α1Change + α2 Ln(Volume10) + α3Volatility10 + α4 Ln(Volume3) + α5Volatility3 + ε (2) Ln(TotalDepth3) = α0 + α1Change + α2 Ln(Volume10) + α3Volatility10 + α4 Ln(Volume3) + α5Volatility3 + ε (3) where the dependent variables represent the bid-ask spread, the logarithm of best depth and the logarithm of total depth, respectively. Change is again a dummy variable assigned the value of 1 if the observation is taken from the post-event sample, or 0 otherwise. Ln(Volume10) is the logarithm of the average daily traded volume in the 10- year bond futures and Volatility10 is the natural logarithm of the highest traded price divided by the lowest traded price for each day in the 10-Year bond futures. Ln(Volume3) is daily traded volume in the 3-Year bond futures. Volatility3 represents volatility in the 3-Year bond futures. The above regressions control for contract-specific and market-wide impacts as it includes the control variables traded volume and volatility for both the 3- Year bond futures and its substitute contract the 10-Year bond futures. In equation (1) the bid-ask spread per minimum tick as opposed to the absolute bid-ask spread is used as the dependent variable for reasons stated above. These three regressions are re-estimated using the 10-Year bond futures as dependent variables. The regression results presented in Table 3 controls for broad market movements using the Chordia, Roll and Subrahmanyam (2000) specification. In Panel A, the results for the bid-ask spread regression for the 3-Year Bond futures shows a significant decline in bid- ask spreads as a function of the minimum tick. A similar result is found for the night trading session. The results for the 10-Year bond futures show a negative coefficient on the dummy variable, indicating that spreads per minimum tick narrowed over the period. The size of the coefficient however was smaller relative to the 3-Year bond and was only significant at the 5 percent level. For night trading the dummy coefficient is positive and statistically significant. As reported in Panel B and C of Table 3, best depth and total visible depth for the 3-Year futures contracts has grown significantly in the post-period after controlling for depth in the 10-Year bond contract. These results are consistent across the night time period. The regressions for the control contract show a reduction in best depth after the change in minimum tick, though this decline is statistically insignificant for both day and night trading. Total depth shows no significant change during the day and a significant increase overnight. This suggests the change in tick size and not market-wide improvements are driving the changes in spreads and depth for the 3-Year bond futures. Similar conclusions are drawn from the results of the combined regression specification of Harris (1994) and Chordia et al., (2000). Panel A of Table 4 shows that after controlling for volatility and volume for both the 3-Year and 10-Year futures contracts, bid-ask spreads as a function of the minimum tick experience a statistically significant decline after the change in tick size for both day and night trading. For the 10-Year bond futures, the dummy coefficient is significantly negative for both day and night trading as well. The regression results in Panel B and C of Table 4 show a significant improvement in both the best and total depth levels after the increase in minimum tick. The coefficient on the change dummy variable is significantly positive for both day and night trading. This result is specific to the 3-Year bond futures, with the coefficient on the change dummy variable for the 10-Year bond contract being insignificant for both day and night trading, except for total depth in overnight trading which exhibited a significant increase. The change in depth is isolated to the 3-Year bond futures contract. As a robustness test, the above regressions were re-estimated using the period of 13 May, 2008 to 13 August, 2008 as the pre-event sample period. Confirming the prior results, regression results in Table 5 and 6 show a decline in bid-ask spreads as a function of the minimum tick after the change in tick size for the 3-Year bond futures, suggesting the increase tick size has resulted in more traders quoting at the minimum tick. Quoted depth for the 3-Year bond futures continue to increase after the tick change when controlling for seasonality in trading, while depth for the 10-Year contract declines. Interestingly, the results reveal that total depth in overnight trading for the 3-Year bond futures exhibited no real change, suggesting the tick size had no apparent affect. However, the results show a decline in total depth on overnight trading for the 10-Year bond futures, indicating total depth did improve relative to market conditions. 4 5 4 The increase in the minimum tick coincided with a reduction in the visibility of the order book in the 3- Year bond futures from 5 to 3 price levels. This change in transparency could affect the impact of the increase in the tick size on liquidity. However, in a study of order-book transparency and market quality on the SFE, Alampieski and Lepone (2006) report no change in market quality when the SFE increased order- book visibility from 3 to 5 levels for the 10-Year bond futures. Given the high substitutability between the 3-Year and 10-Year bond futures, it is therefore considered unlikely that the change in transparency in the 3-Year bonds has had an impact on liquidity. 5 Another factor that could impact liquidity in the 3-Year contract is changes in the cash rate by the Reserve Bank of Australia. To examine whether monetary policy has had an impact on liquidity, the regressions 2.3 Execution Costs The results in the previous section show a reduction in spreads per minimum tick and an increase in absolute bid-ask spreads and quoted depth for the 3-Year bond futures. To provide a more accurate assessment of the change in liquidity after the increase in minimum tick (i.e whether the change in bid-ask spreads dominates the change in quoted depth), we calculate the price impact of executing orders. To calculate price impact, each trade is classified into four mutually exclusive quartiles based on trade size. The first quartile contains the smallest 25% of trade sizes and the fourth quartile contains the largest 25% of trade sizes. The price impact of each trade is measured by calculating the return from the price 10 trades prior to the trade price. The absolute value of each price impact is recorded to incorporate both buy and sell trades. This is averaged across each day (night) and then across each sample period. Price impacts are reported in basis points. Results of the price impact analysis are reported in Table 7. Examining the price impact results for the 3-Year bond, there is no significant change in execution costs across all four quartiles for day and night trading. Execution costs for the first quartile during day trading averaged 0.5793 basis points before the tick increase and 0.5836 basis points after, an insignificant change. A similar result is shown for the fourth quartile, showing an insignificant increase from 0.6350 to 0.6585 basis points. These results differ from Alampieski and Lepone (2009) who report a reduction in execution costs after the tick were re-estimated including the daily settlement price of the 30-Day interbank bill rate. All estimated coefficients on the interest rate variable were insignificant. size reduction on the SFE and indicate that the increase in minimum tick has not had an impact on liquidity. Broad-market movements may be causing execution costs for the 3-Year bond to remain unchanged, with an examination of the 10-Year bond suggesting this is the case. As shown in Panel C of Table 7, there is a reduction in execution costs across all four quartiles for day and night trading for the 10-Year bond. Furthermore, this change is significant for 3 out of the 4 quartiles, indicating execution costs for the 3-Year bond increased relative to market conditions. To further test whether execution costs for the 3-Year bond are driven by market conditions, we estimate price impacts for the sample period 13 May, 2008 to 13 August, 2008. As shown in Table 8, there is a significant increase in price impacts across the four quartiles for both day and night trading for the 3-Year bond. For the 10-Year bond, price impacts are significantly reduced for day trading but are higher for night trading, suggesting market conditions cannot explain the increased price impacts for the 3-Year bond. As the cost of executing trades is significantly higher, we conclude that the increase in the minimum tick has lead to a significant reduction in liquidity. To determine the reason for the increase in execution costs, we calculate the average trade size for the 3-Year bond. The average trade size during the day for the 13 May, 2008 to 13 August, 2008 period was 32 contracts and for the period of 10 February, 2009 to 10 May, 2009 it was 22 contracts. Best depth levels in contrast were 531 and 291 contracts respectively for these two sample periods. This suggests that the increase in spreads dominates the increase in quoted depth as there was already sufficient depth prior to the tick increase to accommodate trading. 3. Conclusion This paper examines the impact of a tick size change on market quality for the Sydney Futures Exchange. Several studies have examined the impact of a tick size reduction on liquidity. The literature for both equities and futures markets provide mixed results. A reduction in the tick size is associated with lower spreads and quoted depth. As these changes have conflicting effects on liquidity, certain studies attribute the change in spreads and depth as indicative of an improvement in liquidity, while other studies conclude a reduction in overall liquidity. This paper is the first to examine the impact of an increase in the tick size under a futures market setting. Results indicate that the change in minimum tick is associated with an increase in bid-ask spreads and depth at the best quotes and throughout the limit order- book. Bid-ask spreads per minimum tick decline after the change, suggesting an increasing proportion of trades quoted at the competitive spread. Despite a reduction spreads per minimum tick, execution costs are higher after the change, with the increase in spreads dominating the higher depth levels. This is attributed to depth at the best levels in the order book before the tick size change being sufficient to accommodate trading. These results hold in overnight trading and are robust to market-wide factors. We conclude that an increase in tick size has a negative impact on liquidity under a futures market setting. Appendix The Sydney Futures Exchange is the largest futures exchange in the Asia-Pacific Region. Trading on the SFE operates through a fully automated electronic limit order book called SYCOM. The two main trader types, local participants and full participants, enter orders directly into the order-book with trades taking place based on price and time precedence rules. The 3-Year bond futures and 10-Year bond futures follow a quarterly expiration cycle. The contracts start on the 15th of December, March, June or September, with settlement occurring 3 days before expiration. Both bonds each have face values of AUD 100,000 and are quoted on a “100-yeild” basis (yield deducted from an index of 100.00). The trading hours for both contracts are between 8:30 and 16:30 hours for daytime trading and 17:10 and 7:00 hours for night time trading during US daylight savings time.6 The 3- Year bond futures has a minimum tick of 0.5 basis points and pre-trade transparency of 5 levels either side of the limit order-book before the 11th of May, 2009 and a minimum tick of 1 basis points after that date and pre-trade transparency of 3 levels either side of the limit order-book. The 10-Year bond futures has a minimum tick of 0.5 basis points and pre-trade transparency of 5 levels either side of the limit order-book. Traders can view in real time prices and order volume on each side of the order book and the traded volume and price of each trade that occurs. Trading identity however is anonymous as broker mnemonics are not visible. 6 Trading is between 17:10 and 7:10 hours for night time trading during US non daylight savings time. References Ahn, H., C.Q. Cao and H. Choe, 1996. Tick size, spread and volume, Journal of Financial Intermediation 5, 2-22. Ahn, H., C.Q. Cao and H. Choe, 1998. Decimalization and competition among stock markets: Evidence from the Toronto Stock Exchange cross-listed securities, Journal of Financial Markets 1, 51-87. Alampieski, K. and A. Lepone, 2009. Impact of a tick size reduction on liquidity: evidence from the Sydney Futures Exchange, Accounting and Finance 49, 1-20. Bacidore, J., 1997. The impact of decimalization on market quality: An empirical investigation of the Toronto Stock Exchange, Journal of Financial Intermediation 9, 92- 120. Bessimbinder, H., 2000. Tick size, spreads, and liquidity: An analysis of Nasdaq securities trading near ten dollars, Journal of Financial Intermediation 9, 213-239. Bortoli, L., Frino A, E. Jarnecic and D. Johnstone, 2006. Limit order book transparency, execution risk, and market liquidity: Evidence from the Sydney Futures Exchange, Journal of Futures Markets 26, 1147-1168 Bourghelle, D. and F. Declerck, 2004. Why markets should not necessarily reduce the tick size, Journal of Banking and Finance 28, 373–398. Chakravarty, S., V. Panchapagesan and R. Wood, 2005. Did decimalization hurt institutional investors?, Journal of Financial Markets 8, 400-420. Chordia, T., R. Roll and A. Subrahmanyam, 2000. Commonality in liquidity, Journal of Financial Economics 56, 3-28. Cordella, T. and T. Foucault, 1999. Minimum Price Variations, Time Priority, and Quote Dynamics, Journal of Financial Intermediation 8, 141-173. Frino, A., D. Gerace and A. Lepone, 2008. Liquidity in auction and specialist market structures: Evidence from the Italian bourse, Journal of Banking & Finance 32, 2581- 2588. Frino, A. and M. McKenzie, 2002. The pricing of stock index futures spreads at contract expiration, Journal of Futures Markets 22, 451 - 469. ap Gwilym, O., I. McManus and S. Thomas, 2005. Fractional versus decimal pricing: Evidence from the UK Long Gilt Futures market, The Journal of Futures Markets 25, 429-442. Goldstein, M.A. and K.A. Kavajecz, 2000. Eighths, sixteenths, and market depth: Changes in tick size and liquidity provision on the NYSE?, Journal of Financial Economics 56, 125-149. Harris, L.E., 1994. Minimum price variations, discrete bid-ask spreads, and quotation sizes, The Review of Financial Studies 7, 149-178. Lee, C., B. Mucklow and M. Ready, 1993. Spreads, depths, and the impact of earnings information: An intraday analysis, The Review of Financial Studies 6, 345-374. Ricker, J.P., (1998). 'Breaking the eighth: Sixteenths on the New York Stock Exchange '. working paper. 1730 Filbert Street No. 105, San Francisco. CA 94123. Table 1 Descriptive Statistics Descriptive statistics are presented for measures of market liquidity surrounding the increase in minimum tick in the 3-Year bond futures. The tick size of the 3-Year bond futures was increased from a half to one basis point on 11th May 2009. The pre-event sample period extends from 10 February, 2009 to 10 May, 2009. The post-event sample period extends from 13 May, 2009 to 13 August, 2009. Bid-ask spreads and depth are sampled every 5 minutes (15 minutes) and then averaged for each day (night). Bid-Ask Spread is the best ask price minus the best bid price in contract points. BAS is calculated as the bid-ask spread divided by the minimum tick. Best Depth is the aggregate order volume at the best bid and best ask price. Total depth is the aggregate order volume throughout the limit-order book. Volatility is the natural logarithm of the highest traded price divided by the lowest traded price for each day. Volume is the average daily traded volume. Night trading results are presented in parentheses. Bid-Ask Spread BAS Best Depth Total Depth Volatility Volume Panel A - Pre-period 3-Year 0.0053 1.055 291 1832 0.1027 39665 (0.0089) (1.774) (134) (1008) (0.0902) (10799) 10-Year 0.0054 1.072 110 654 0.1002 17029 (0.0096) (1.912) (50) (344) (0.0946) (4378) Panel B - Post-period 3-Year 0.0102 1.015 1179 4686 0.1008 51461 (0.0119) (1.190) (389) (1610) (0.0962) (16180) 10-Year 0.0052 1.034 148 865 0.0921 15356 (0.0089) (1.780) (60) (387) (0.0972) (5360) Panel C - Change from pre-period to post-period 3-Year 0.0050** -0.0314** 959** 2853** -0.0018 11796** (0.0040)** (-0.5066)** (289)** (602)** (0.0060) (5381)** 10-Year -0.0002** -0.0379** 37** 210** -0.0018 -1673 (-0.0007)* (-0.1308)* (10)** (44)* (0.003) (981)** ** Indicates statistical significance at the 0.01 level * Indicates statistical significance at the 0.05 level Table 2 Descriptive Statistics – Seasonal Control Descriptive statistics are presented for measures of market liquidity surrounding the increase in minimum tick in the 3-Year bond futures. The tick size of the 3-Year bond futures was increased from half to a full- basis point on 11th May 2009. The pre-event sample period extends from 13 May, 2008 to 13 August, 2008. The post-event sample period extends from 13 May, 2009 to 13 August, 2009. Bid-ask spreads and depth are sampled every 5 minutes (15 minutes) and then averaged for each day (night). Bid-Ask Spread is the best ask price minus the best bid price in contract points. BAS is calculated as the bid-ask spread divided by the minimum tick. Best Depth is the aggregate order volume at the best bid and best ask price. Total depth is the aggregate order volume throughout the limit-order book. Volatility is the natural logarithm of the highest traded price divided by the lowest traded price for each day. Volume is the average daily traded volume. Night trading results are presented in parentheses. Bid-Ask Spread BAS Best Depth Total Depth Volatility Volume Panel A - Pre-period 3-Year 0.0052 1.043 531 3471 0.0860 48939 (0.0080) (1.600) (214) (1645) (0.0785) (14586) 10-Year 0.0053 1.059 192 1227 0.0860 19662 (0.0081) (1.625) (85) (514) (0.0832) (7765) Panel B - Post-period 3-Year 0.0102 1.015 1179 4686 0.1008 51461 (0.0119) (1.190) (389) (1610) (0.0962) (16180) 10-Year 0.0052 1.034 148 865 0.1008 15356 (0.0090) (1.791) (60) (389) (0.0972) (5360) Panel C - Change from pre-period to post-period 3-Year 0.0050** -0.0215** 724** 1215** 0.0150* 2521 (0.0040)** (-0.3339)** (210)** (-35) (0.018)** (1595) 10-Year -0.0001** -0.0250** -44** -362** 0.0150* -4307** (0.0008)** (0.1660)** (-26)** (-125)** (0.014)** (-2405)** ** Indicates statistical significance at the 0.01 level * Indicates statistical significance at the 0.05 level Table 3 Market wide regressions This table reports the regression results of spreads and depth around the move to full-basis point trading in the 3-Year contract. The pre-event sample period extends from 10 February, 2009 to 10 May, 2009. The post-event sample period extends from 13 May, 2009 to 13 August, 2009. The following regression models are estimated: BASi = α0 + α1Change + α2 ControlBASi + ε Ln(BestDepthi) = α0 + α1Change + α2Ln(ControlBestDepthi) + ε Ln(TotalDepthi) = α0 + α1Change + α2Ln(ControlTotalDepthi) + ε where BASi is the bid-ask spread per minimum tick for each contract. Ln(BestDepthi) is the logarithm of the aggregate order volume at the best bid and best ask price. Ln(TotalDepthi) is the logarithm of the aggregate order volume throughout the limit-order book. The variable Change is a dummy variable assigned the value of 1 if the observation is taken from the post-event sample, or 0 otherwise. Night trading results are presented in parentheses. All t statistics are adjusted for heteroscedasticity and autocorrelation using the procedure by Newey and West (1987). Intercept Change Control(10yr) Control(3yr) R2 Panel A: Bid-Ask Spreads BAS3 0.7462** -0.0295** 0.2885** - 0.4865 (1.184)** (-0.5433)** (0.3088)** - (0.7157) BAS10 0.5071** -0.0162* - - 0.3480 (0.8781)** (0.2090)* - (0.5822)** (0.2064) Panel B: Best Depth Ln(BestDepth3) 3.881** 1.378** 0.3502** - 0.9239 (4.021)** (1.070)** (0.1511) - (0.7834) Ln(BestDepth10) 2.289** -0.2372 - 0.3982** 0.4295 (2.781)** (0.0445) - (0.1547) (0.1232) Panel C: Total Depth Ln(TotalDepth3) 5.305** 0.8636** 0.3321** - 0.8166 (7.213)** (0.4463)** (-0.0682) - (0.3909) Ln(TotalDepth10) 3.763** -0.0273 - 0.3530** 0.3524 (6.282)** (0.1898)* - (-0.0866) (0.0487) ** Indicates statistical significance at the 0.01 level * Indicates statistical significance at the 0.05 level Table 4 Combined Regressions This table reports the regression results of spreads and depth around the move to full-basis point trading in the 3-Year contract. The pre-event sample period extends from 10 February, 2009 to 10 May, 2009. The post-event sample period extends from 13 May, 2009 to 13 August, 2009. The following regression models are estimated: BASi = α0 + α1Change + α2 Ln(Volume10) + α3Volatility10 + α4 Ln(Volume3) + α5Volatility3 Ln(BestDepthi) = α0 + α1Change + α2 Ln(Volume10) + α3Volatility10 + α4 Ln(Volume3) + α5Volatility3 Ln(TotalDepthi) = α0 + α1Change + α2 Ln(Volume10) + α3Volatility10 + α4 Ln(Volume3) + α5Volatility3 where BASi is the bid-ask spread per minimum tick for each contract. Ln(BestDepthi) is the logarithm of the aggregate order volume at the best bid and best ask price. Ln(TotalDepthi) is the logarithm of the aggregate order volume throughout the limit-order book. The variable Change is a dummy variable assigned the value of 1 if the observation is taken from the post-event sample, or 0 otherwise. Ln(Volume3) is the logarithm of the average daily traded volume in the 3-year bond futures. Volatility3 is the natural logarithm of the highest traded price divided by the lowest traded price for each day in the 3-Year bond futures. Ln(Volume10) is daily traded volume in the 10-Year bond futures. Volatility10 represents volatility in the 10-Year bond futures. Regression results are presented separately for the 3-Year and 10-Year bond futures contracts. Night trading results are presented in parentheses. All t statistics are adjusted for heteroscedasticity and autocorrelation using the procedure by Newey and West (1987). Intercept Change Ln(Volume10) Volatility10 Ln(Volume3) Volatility3 R2 Panel A: Bid-Ask Spreads BAS3 1.153** -0.0421** -0.0145 0.0989 0.0028 0.0384 0.4173 (2.922)** (-0.5269)** (0.0282) (0.6776) (-0.1538) (-0.3333) (0.6686) BAS10 1.158** -0.0455** -0.0279* -0.0827 0.0169 0.1474 0.2876 (2.469)** (-0.1377)* (-0.2424)** (0.7058) (0.1620) (-1.179) (0.0926) Panel B: Best Depth Ln(Best Depth3) 2.967** 1.412** -0.0583 0.1949 0.3129** -2.676** 0.9233 (4.311)** (1.051)** (-0.3067)** (0.5873) (0.3105)* (-1.502) (0.8048) Ln(BestDepth10) 1.160 0.4572** 0.5414** -0.2494 0.3446* -3.213 0.6030 (2.166)** (0.2030)** (0.3496)** (-1.627) (0.3834) (-2.304) (0.5002) Panel C:Total Depth Ln(TotalDepth3) 4.532** 0.8667** -0.0435 0.1580 0.3446** -0.8724** 0.8298 (5.435)** (0.3449)** (-0.2278)** (-0.4510) (0.3834)** (0.7213) (0.5002) Ln(TotalDepth10) 3.185** 0.3935** 0.4945** -0.3537 -0.1364* -1.188 0.5191 (5.849)** (0.2053)** (0.3640)** (-2.085) (-0.3397)** (1.648) (0.2003) ** Indicates statistical significance at the 0.01 level * Indicates statistical significance at the 0.05 level Table 5 Market wide regressions – Seasonal Control This table reports the regression results of spreads and depth around the move to full-basis point trading in the 3-Year contract. The pre-event sample period extends from 13 May, 2008 to 13 August, 2008. The post- event sample period extends from 13 May, 2009 to 13 August, 2009. The following regression models are estimated: BASi = α0 + α1Change + α2 ControlBASi + ε Ln(BestDepthi) = α0 + α1Change + α2Ln(ControlBestDepthi) + ε Ln(TotalDepthi) = α0 + α1Change + α2Ln(ControlTotalDepthi) + ε where BASi is the bid-ask spread per minimum tick for each contract. Ln(BestDepthi) is the logarithm of the aggregate order volume at the best bid and best ask price. Ln(TotalDepthi) is the logarithm of the aggregate order volume throughout the limit-order book. The variable Change is a dummy variable assigned the value of 1 if the observation is taken from the post-event sample, or 0 otherwise. All t statistics are adjusted for heteroscedasticity and autocorrelation using the procedure by Newey and West (1987). Intercept Change Control(10yr) Control(3yr) R2 Panel A: Bid-Ask Spreads BAS3 0.9004** -0.0250** 0.1350** - 0.3717 (1.278)** (-0.4416)** (0.1989)** - (0.5023) BAS10 0.7763** -0.0173** - 0.2711* 0.1941 (1.189)** (0.2669)** - (0.2727)** (0.1231) Panel B: Best Depth Ln(BestDepth3) 3.477** 1.011** 0.5102** - 0.7707 (3.838)** (0.7771)** (0.2797)* - (0.5674) Ln(BestDepth10) 2.881** -0.5580** - 0.3591** 0.3451 (2.744)** (-0.5388)** - (0.2646)* (0.3035) Panel C: Total Depth Ln(TotalDepth3) 5.325** 0.4616** 0.3892** - 0.3559 (5.770)** (0.0325) (0.2496)* - (0.0430) Ln(TotalDepth10) 4.756** -0.4414** - 0.2840** 0.4468 (4.730)** (-0.2859)** - (0.1928)* (0.2665) ** Indicates statistical significance at the 0.01 level * Indicates statistical significance at the 0.05 level Table 6 Combined Regressions – Seasonal Control This table reports regression results of spreads and depth around the move to full-basis point trading in the 3-Year contract. The pre-event sample period extends from 13 May, 2008 to 13 August, 2008. The post-event sample period extends from 13 May, 2009 to 13 August, 2009. The following regression models are estimated: BASi = α0 + α1Change + α2 Ln(Volume10) + α3Volatility10 + α4 Ln(Volume3) + α5Volatility3 Ln(BestDepthi) = α0 + α1Change + α2 Ln(Volume10) + α3Volatility10 + α4 Ln(Volume3) + α5Volatility3 Ln(TotalDepthi) = α0 + α1Change + α2 Ln(Volume10) + α3Volatility10 + α4 Ln(Volume3) + α5Volatility3 where BASi is the bid-ask spread per minimum tick for each contract. Ln(BestDepthi) is the logarithm of the aggregate order volume at the best bid and best ask price. Ln(TotalDepthi) is the logarithm of the aggregate order volume throughout the limit-order book. The variable Change is a dummy variable assigned the value of 1 if the observation is taken from the post-event sample, or 0 otherwise. Ln(Volume3) is the logarithm of the average daily traded volume in the 3-year bond futures. Volatility3 is the natural logarithm of the highest traded price divided by the lowest traded price for each day in the 3-Year bond futures. Ln(Volume10) is daily traded volume in the 10-Year bond futures. Volatility10 represents volatility in the 10-Year bond futures. Regression results are presented separately for the 3-Year and 10-Year bond futures contracts. Night trading results are presented in parentheses. All t statistics are adjusted for heteroscedasticity and autocorrelation using the procedure by Newey and West (1987). Intercept Change Ln(Volume10) Volatility10 Ln(Volume3) Volatility3 R2 Panel A: Bid-Ask Spreads BAS3 1.171** -0.0285** -0.0008 -0.1521* -0.0116* 0.1789** 0.3710 (1.544)* (-0.3893)** (0.1014) (1.772) (-0.1058) (0.1870) (0.5060) BAS10 1.136** -0.0301** -0.0135 -0.0613 0.0046 0.1370 0.1804 (2.041)** (0.0911) (-0.1673)* (1.861) (0.1133) (-2.060) (0.1014) Panel B: Best Depth Ln(Best Depth3) 1.755 0.8727** -0.0668 -0.0002 0.4870** -2.980** 0.1804 (4.057)** (0.5921)** (-0.2426)* (-0.7255) (0.3471)* (-2.078) (0.1014) Ln(BestDepth10) 2.558** -0.1040* 0.4294* 0.5729 0.4151* -1.435* 0.4046 (2.522)** (-0.1888)** (0.3526)** (-3.096)* (0.3506) (0.9440) (0.3415) Panel C:Total Depth Ln(TotalDepth3) 4.351** 0.3273** -0.0440 -0.3102 0.4151** -3.147** 0.434547 (6.102)** (-0.0814) (-0.1992) (-0.9114) (0.3506)** (-3.595)** (0.1580) Ln(TotalDepth10) 5.630** -0.2330** 0.3562** -0.2891 -0.1888* -0.5096 0.4843 (5.134)** (-0.1476)* (0.2847)** (-3.724)* (-0.1383) (1.293) (0.3082) ** Indicates statistical significance at the 0.01 level * Indicates statistical significance at the 0.05 level Table 7 Execution Costs Price impact results are presented for the 3-Year and 10-Year bonds both before and after the change in tick size for the 3-Year bond. The pre-event sample period extends from 10 February, 2009 to 10 May, 2009. The post-event sample period extends from 13 May, 2009 to 13 August, 2009. The price impact of each trade is measured as the return from the transaction price ten trades prior to the trade price. The absolute value of each price impact is recorded to incorporate both buy and sell trades. This is averaged across each day (night) and then across each sample period. Each trade is classified into four mutually exclusive quartiles based on trade size. The first quartile contains the smallest 25% of trade-sizes and the fourth quartile contains the largest 25% of trade-sizes. Price impact is reported in basis points. Night trading results are in parentheses. Quartile 1 Quartile 2 Quartile 3 Quartile 4 Panel A - Pre-period 3-Year 0.5793 0.5776 0.6195 0.635 (1.058) (1.118) (1.118) (1.230) 10-Year 0.5865 0.6016 0.5960 0.5495 (1.213) (1.221) (1.188) (1.223) Panel B - Post-period 3-Year 0.5836 0.5234 0.6595 0.6585 (1.040) (1.064) (1.170) (1.200) 10-Year 0.5535 0.5387 0.5349 0.5086 (1.087) (1.064) (1.082) (1.032) Panel C - Change from pre-period to post-period 3-Year 0.0043 -0.0542 0.0400* 0.0235 (-0.0178) (-0.0545) (0.0521) (-0.0292) 10-Year -0.0330 -0.0629** -0.0611** -0.0409* (-0.1258)* (-0.1565)** (-0.1060) (-0.1911)** ** Indicates statistical significance at the 0.01 level * Indicates statistical significance at the 0.05 level Table 8 Execution Costs – Seasonal Control Price impact results are presented for the 3-Year and 10-Year bonds both before and after the change in tick size for the 3-Year bond. The pre-event sample period extends from 13 May, 2008 to 13 August, 2008. The post-event sample period extends from 13 May, 2009 to 13 August, 2009. The price impact of each trade is measured as the return from the transaction price ten trades prior to the trade price. The absolute value of each price impact is recorded to incorporate both buy and sell trades. This is averaged across each day (night) and then across each sample period. Each trade is classified into four mutually exclusive quartiles based on trade size. The first quartile contains the smallest 25% of trade-sizes and the fourth quartile contains the largest 25% of trade-sizes. Price impact is reported in basis points. Night trading results are in parentheses. Quartile 1 Quartile 2 Quartile 3 Quartile 4 Panel A - Pre-period 3-Year 0.5011 0.4952 0.5732 0.5744 (0.9403) (0.8967) (0.9984) (0.9574) 10-Year 0.6127 0.5742 0.5711 0.5703 (0.9215) (0.8983) (0.9297) (0.8782) Panel B - Post-period 3-Year 0.5836 0.5234 0.6595 0.6585 (1.040) (1.064) (1.170) (1.200) 10-Year 0.5535 0.5387 0.5349 0.5086 (1.087) (1.064) (1.082) (1.032) Panel C - Change from pre-period to post-period 3-Year 0.0825** 0.0282 0.0863** -0.0841** (0.0994)* (0.1669)** (0.1715)** (0.243)** 10-Year -0.0592** -0.0355* -0.0362* -0.0617** (0.1652)** (0.1661)** (0.1518)** (0.1539)** ** Indicates statistical significance at the 0.01 level * Indicates statistical significance at the 0.05 level

DOCUMENT INFO

Shared By:

Categories:

Tags:
market quality, order book, limit order, bond futures, bid-ask spread, futures price, 10 Year, price volatility, price levels, futures contract

Stats:

views: | 6 |

posted: | 4/13/2010 |

language: | English |

pages: | 31 |

OTHER DOCS BY accinent

How are you planning on using Docstoc?
BUSINESS
PERSONAL

By registering with docstoc.com you agree to our
privacy policy and
terms of service, and to receive content and offer notifications.

Docstoc is the premier online destination to start and grow small businesses. It hosts the best quality and widest selection of professional documents (over 20 million) and resources including expert videos, articles and productivity tools to make every small business better.

Search or Browse for any specific document or resource you need for your business. Or explore our curated resources for Starting a Business, Growing a Business or for Professional Development.

Feel free to Contact Us with any questions you might have.