Fragmentation in Asset Markets: the price discovery implications of competitive fragmentation in equity and cryptocurrency markets.

  • Jaroslaw Majtyka

Student thesis: Doctoral Thesis

Abstract

This thesis examines fragmentation in asset markets. A taxonomical framework is developed to characterise the motivational factors behind the innovations that result in market fragmentation. Drawing on this framework and the relevant literature this thesis empirically examines the relationship between fragmentation and one of the motivational factors, reductions in information asymmetry, otherwise known as price discovery, within equity and cryptocurrency markets. Rational expectations theory is a cornerstone of the efficient market hypothesis and the basis of models proposed by Kyle (1985) and Glosten and Milgrom (1985) and their derivatives. The theory suggests that investors use available information, past experiences, and their own rational expectations to make decisions that influence asset valuation. Tests regarding whether market fragmentation contributes to the dispersal of fundamental price adjusting information are imperative in determining the role competition amongst exchanges plays in the market’s ability to maintain prices at efficient levels. If prices are to remain efficient, and information asymmetry kept to a minimum, exchanges must incorporate relevant information across all trading platforms. Such a process is made more difficult and time-consuming as information becomes more decentralised or fragmented. The term ‘market fragmentation’ is used heterogeneously within the literature to refer to events that segregate market participants from one another or when prices across trading platforms deviate from the fundamental equilibrium. Competitive fragmentation describes events that place investors trading a common asset into separate trading pools due to competition among exchanges. Fragmentation based on investor type differentiates between investors based on characteristics such as geographic location and investor class. Substitutionary fragmentation occurs when markets present investors with the option to purchase derivative products or assets that are considered direct substitutes for existing products. Financial fragmentation refers to periods when assets deviate from their fundamental value across a subset of exchanges. The first study adapts and expands upon the taxonomies presented by Avlonitis, Papastathopoulou, and Gounaris (2001) and Tufano (1989). It establishes that most innovations that lead to competitive, investor based and substitutionary fragmentation, are motived by a desire to reduce transaction costs, with reductions in information asymmetry playing a supporting role. These innovations are often modifications or extensions to existing services. Modern motivational factors exert greater influence over recent fragmenting innovations and represent the formation of new products or trading methods. Technological shocks and globalisation are most responsible for fragmenting events involving dark pools, cryptocurrencies, and high-frequency trading. The second study investigates the role of order book transparency on the relationship between price discovery and market fragmentation. It utilises Hasbrouck’s (1995) information share and Gonzalo and Granger’s (1995) component share to measure price discovery across 120 stocks from six European countries (2008-2016). Panel-regression results support existing theory on price discovery in equity markets in that displayed (lit) order book prices contain substantially more information than non-displayed (dark) prices (Zhu, 2014). Dark market share coefficients, which measure competitive fragmentation between lit and dark exchanges, provide additional support that dark transactions are substantially less informed than lit transactions. Informed investors are discouraged from trading in dark pools due to high levels of non-execution risk. Fragmentation is associated with greater adverse selection risk in quoting exchanges as informed investors use their informational advantage to supply liquidity (Rindi, 2008). Mid-quotes on lit exchanges are also more informative than lit prices (Bloomfield, O’hara, & Saar, 2005; Boulatov & George, 2013). Increases in fragmentation among quoting (lit) exchanges lead to a decrease in the informativeness of lit trades versus dark trades in both the primary and consolidated lit markets. The informativeness of exchange trades as compared to quotes deteriorates with greater competitive fragmentation across lit exchanges suggesting that lit fragmentation is associated with higher levels of adverse selection. The negative relationship between price discovery and volatility further supports this claim and is consistent with the notion that the most profitable uninformed trades are being ‘skimmed’ by informed liquidity providers (Bessembinder & Kaufman, 1997; Easley, Kiefer, & O'Hara, 1996). The final study applied existing research on competitive fragmentation and price discovery to cryptocurrency markets in order to test its applicability to a new asset class. Bitcoin (BTC) transaction and order book data is collected across six exchanges for both United States Dollar (USD - $) and Euro (€) order books (2017-2019). A panel-regression model on a multivariate version of Hasbrouck’s (1995) information share is employed. Results confirm previous findings that market share has a positive relationship with the informativeness of exchange prices (Madhavan, 1995). Consistent with the previous study, this is attributed to informed investors migrating to competing exchanges to better conceal and profit on their superior information. This, in turn, increases events of information asymmetry as exchange prices become more informative and dispersed across an increasing number of exchanges. The results contained in this thesis suggest that competitive fragmentation is in part, motivated by the intention to reduce information asymmetry. However, in practice information asymmetry increases. Competing exchanges attract a disproportionate amount of uninformed trading, though some informed investors follow the uninformed to competing exchanges in order to capitalise on their informational advantage. Finally, consistencies are found between equity and cryptocurrency markets suggesting that theories developed around traditional asset classes are transferable to the newly formed market for cryptocurrencies.
Date of Award2021
Original languageEnglish
SupervisorKeith Duncan (Supervisor) & Simone Kelly (Supervisor)

Cite this

'