Exchange Competition, Fragmentation and the Destruction of Liquidity in Markets that Operate Centralised Electronic Limit Order Books

Author Name: 
Alex Frino

In the early 1980s and 1990s US regulators commenced promoting competition that fragmented trading on US exchanges in an effort to put competitive pressure on participants on those exchanges: specifically specialists and dealers.  The main rationale of US regulators was that these exchange participants enjoyed various forms of market power and they used this market power to extract economic rents from investors by quoting excessively wide bid-ask spreads.  The regulators reasoned that any upward pressure on bid-ask spreads resulting from fragmentation would be more than offset from the downward pressure on bid ask spreads through competition foisted on specialists and dealers, thereby reducing bid ask spreads and the cost of trading.  The measures advocated by the regulators included the promotion of competition between exchanges and regional exchanges as well as upstairs or off-market trading.  More recently, the opportunity to trade upstairs and technological innovation has spawned dark pools, which continue to fragment US markets.  The initial effects of the competition and fragmentation in US markets on bid-ask spreads (decades ago) has been documented in numerous academic studies.  These studies have confirmed that competing exchanges and upstairs trading historically reduced on-exchange bid-ask spreads and therefore reduced transaction costs thereby enhancing liquidity.[1][2]

Roll the clock forward two decades, and international regulators have commenced promoting the same sorts of measures in markets where there is a centralised exchange that operates an electronic limit order book, including Australia and the England.  In both these markets competing exchanges have been introduced and dark pools are beginning to proliferate.  Well-meaning regulators estimate that these measures provide competition for the exchanges thereby forcing them to reduce their fees and in turn this reduces overall transaction costs and improves market liquidity.  Unfortunately, this rationale is flawed.  Competition and the fragmentation that it causes in these markets is unlikely to reduce transaction costs and poses a threat to liquidity. While on the face of it, the measures seem the same as those promoted by US regulators decades ago, and one would presume that they would have the same effect, there is a subtle but very important difference between the two.  The competition in the US was primarily designed to challenge participants of exchanges, whereas the competition in countries such as Australia and England are purely directed at the exchanges themselves.  To understand the impact of this difference, consider the breakdown in the costs of trading on exchanges.

The Costs of Trading Equities

Table 1 and Diagram 1 below identify the relative size of the 3 major costs of trading in equities markets.  The table sets out estimates of the round-trip cost of trading on the Australian Securities Exchange.  The diagram illustrates that the exchange fee is trivial compared to bid-ask spreads and brokerage fees.  In the Australian market, for example, the average bid-ask spread across the top 200 stocks is 100 times the magnitude of the exchange fee.   Similarly, average brokerage fees in the same stocks are also 100 times the magnitude of the exchange fee.  It is quite clear, therefore, that to introduce competition and fragmentation in the Australian market in order to reduce overall transaction costs is flawed.  While competing exchanges and the promotion of off-market trading may put pressure on exchanges to reduce their fees, at the same time, the fragmentation that results will put upward pressure on the much larger components of transaction costs.  To be clear, even a small increase in bid-ask spreads will completely swamp a small decrease in the exchange fee.  This will increase overall transaction costs and thereby destroy liquidity.  In the following section, I report the results of some analysis which quantifies this for the Australian market.


Diagram 1 – The cost of trading on the Australian Stock Exchange









Table 1 The cost of trading on the Australian Stock Exchange










ASX Annual Report 2011 (whole market)




12 of the largest fund managers for 1st half of 2010 (benchmark portolios)

Bid-Ask Spread



Simple average for ASX 200, based on Reuters data & own calculations for the year ended 30 December, 2011



The Impact of Fragmentation on Bid-Ask Spreads

In 1968 a UCLA academic by the name of Harold Demsetz wrote a seminal paper that spawned an entirely new field of study in finance known as market microstructure.[5]  In the paper, Professor Demsetz identified that bid-ask spreads are a cost of trading, and that they are negatively related to trading activity.  He argued that if traders posting quotes (or limit orders) have to wait longer to trade because trading activity is lower, then they will lower their buying price or increase their selling price in order to compensate themselves from the costs and risks from waiting longer to trade.  Using data from the New York Stock Exchange, he confirmed this finding empirically.  His core result – that bid ask spreads and trading activity are inversely related -is now taken to be virtually a universal law in finance.  It has been demonstrated empirically by literally hundreds of academic papers for many markets around the world including Australia.[6]  This law holds the key to understanding, and indeed estimating the increase in bid-ask spreads, transaction costs and therefore the destruction of liquidity that results from fragmentation of markets which operate centralised electronic limit order books.  Put simply, when a market which operates a centralised limit order book becomes fragmented, all else remaining equal, trading activity is lower than it otherwise would be on that market.  This causes bid-ask spreads to widen thereby destroying liquidity.

Demsetz original model was very similar to the following:

Bid-ask spread = a + b Volume + c Volatility

The model above says that the bid ask spread is a linear function of trading volume and price volatility.  The coefficients b and c determine the impact that trading volume and price volatility, respectively, have on bid-ask spreads and can be estimated using a statistical technique such as regression analysis.  Importantly, if we were to estimate the model above, the coefficient b could be used to forecast the impact on bid-ask spreads from a reduction in trading volume through fragmentation.  The coefficient b, of course, is expected to be negative.


Using daily data for the stocks listed on the Australian Securities Exchange over the period 1 Jan 2009 to 31 December 2011, and data obtained from Reuters (TRTH) and Bloomberg, (2012) estimated the values of a, b and c for each of the stocks in the S&P/ASX 300. [7]  Most importantly, the average estimate for b for this group of stocks is -0.16.  This very important coefficient can be used to estimate the impact of market fragmentation on the bid ask spreads of the stocks in the ASX 200.  For example, if fragmentation on the ASX causes trading activity to fall by a mere 10 percent, then bid ask spreads will increase by approximately 1.5 percent (-0.16 x 0.10).  Given that the average bid-ask spread across the top 200 stocks on the ASX is 0.35 percent (see Table 1), then the increase in the bid ask spread if 20 percent of lit market trading activity fragments is 0.5 basis points (0.015 x 0.35).  This is more than 1.5 times the current exchange fee. 

This analysis suggests that even if the fragmentation of the market caused exchange fees for trading to be completed eliminated on the Australian Securities Exchange that the market will be worse-off as the bid ask spread will rise by many times more than that in a conservative scenario where trading volume declines a mere 10 percent due to fragmentation.  It clearly illustrates the fallacy of the argument that exchange competition and market  fragmentation would improve the costs of trading in markets that operate centralised electronic limit order books such as Australia.  In fact, fragmentation of trading in such markets increases the overall cost of trading and destroys liquidity.

[1] I define liquidity to be the ability to convert securities into cash quickly and cheaply.  Since the bid-ask spread is a cost of immediately or quickly converting a security into cash, then an increase in bid ask spreads is consistent with a decrease in liquidity.

[2][2] These studies include, for example, Cohen, K and R. Conroy, 1990, “An empirical study of the effect of rule 19c3”, Journal of Law and Economics and McInish, T. and R.A. Wood, 1996, “Competition, Fragmentation and Market Quality”, The Industrial Organisation and Regulation of the Securities Industry, National Bureau of Economic Research,.

[3] While this article focuses on the effects of fragmentation on bid-ask spreads, it is also likely that brokerage fees will increase significantly from fragmentation.  Brokers are forced to expand their infrastructure to cope with multiple execution venues, which increases their costs and is likely to increase the fees that they charge.

[4] I would like to thank Professor David Gallagher from the Macquarie Graduate School of Management for providing this estimate based on the transaction data that he has collected from 12 of the largest fund managers in Australia.

[5] Demsetz, H., 1968, “The cost of transacting”, The Quarterly Journal of Economics.

[6] See Aitken, M. and A. Frino, 1996, “The Determinants of Bid-Ask Spreads on the Australian Stock Exchange”, Accounting and Finance.

[7]  The bid ask spread at the end of each trading day is the difference between the best bid and ask price divided by the average (bid and ask) price, volume is the number of shares traded and volatility is the logarithm of the ratio of the highest price traded to the lowest price traded.  The model is estimated in log form so that the coefficients are all elasticities.  For further detail, see Frino, A., V. Mollica and S. Zhang, 2012, “Determinants of Bid-Ask Spreads in Time-Series Analysis: A Note”, Working Paper, Capital Markets Cooperative Research Centre.

About the author :

Alex Frino is Chief Executive Officer of Capital Markets CRC Limited, and also Professor of Finance at the University of Sydney.    He holds a Masters Degree in Finance from the University of Cambridge, England and a PhD in Finance from the University of Sydney, Australia.    Professor Frino specialises in financial market integrity and microstructure.   He has published over 100 articles in leading scholarly journals.   He has previously held positions with CS First Boston Australia and the Sydney Futures Exchange.   Alex Frino was the recipient of the 2005 Fulbright Senior Scholar Award.   He currently serves on the Disciplinary Tribunal of the Australian Securities Exchange.