6 Exploring theoretical basics – scale effects in clearing While the previous chapter delivered a number of quantitative and qualitative analyses of European derivatives clearing costs, this chapter explores and classifies possible scale effects in clearing. The insights provided by Chapters 5 and 6 serve as a basis for the subsequent analysis of the efficiency impact of different network strategies in Chapters 7, 8 and 9. The economic literature distinguishes between demand- and supply- side scale effects. 1 Demand-side scale effects are commonly referred to as network effects. Supply-side scale effects include economies of scale and scope. In the following, these concepts are introduced and applied to clearing. Additionally, evidence for the existence of demand- and supply-side scale effects in clearing is explored. In a first step (section 6.1), demand-side scale effects and their economic implications are analysed. A second step (section 6.2) investigates supply-side effects. This includes insight to clearing houses’ cost structures in order to identify the causes and nature of supply-side scale effects. Finally, this chapter’s findings are summarised (section 6.3). 6.1 Demand-side scale effects There exist various industries in which the utility that a user derives from a given product or service increases with the number of other consumers utilising the same product or ser vice. 2 Inthiscase,theusersoftheprod- uct or service constitute a network. The most prominent example of this 1 Cf. Farrell/Saloner (1986), p. 940; Katz/Shapiro (1986a), p. 824; Bessler (1991), p. 275; Besen/Farrell (1994), p. 118; and Junius (1997), p. 7. Alternatively, scale effects can be differentiated as cost- and revenue-sided. Cf. Berger/Humphrey (1997), p. 9. When applying the economic concept of scale effects to clearing, the analysis is based on the separation of the demand and supply sides, because not all clearing houses act according to the principle of profit maximisation. 2 Cf. Katz/Shapiro (1985), p. 424. See Shy (2001) for a variety of industry examples. 196 Clearing Services for Global Markets EXPLORING THEORETICAL BASICS – SCALE EFFECTS IN CLEARING Demand-Side Scale Effects 6.1 6 Scale effects are inherent to network strategies and can influence transaction costs. This chapter introduces the possible scale effects in clearing. Examines the existence of demand-side scale effects, i.e. network effects, in clearing; conclusions are drawn on the economic implications. Examines the existence of supply-side scale effects, i.e. economies of scale and scope, in clearing. Analyses clearing houses’ cost structures to allow for a more detailed examination of the causes and nature of supply-side scale effects. 6.2 Supply-Side Scale Effects Summary of Findings6.3 ESOPRUPRETPAHC Figure 6.1 Structure of Chapter 6 effect can be found in communication networks, such as the public telephone network. 3 Because the value of membership to one user is positively affected when another user joins and enlarges the network, 4 the corresponding markets are said to exhibit network effects 5 or network externalities. 6 Network effects are classified as scale effects on the demand side, 7 because with a linearly 3 For an early analysis of positive network externalities in communications networks, see Rohlfs (1974), p. 16. 4 As the number of individuals utilising phones increases, so does the number of connection alternatives. In this case, the benefit for individuals is the possibility to communicate with an increased number of other individuals. 5 The term results from the fact that these effects were first recognised and studied in the context of communication networks, such as those associated with the telephone and telex. However, network effects also exist in many other industries that do not employ physical networks. Cf. Katz/Shapiro (1986b), p. 146. In other words, both physical and virtual networks can exhibit network effects. 6 Cf. Katz/Shapiro (1994), p. 94. When deciding to join a network, users do not commonly take into account the positive welfare effect associated with their decision. Users thus do not internalise the welfare effect, which results i n an externality. Network effects are often externalities, but need not be. While network effects formally apply to a more general concept than network externalities, the two terms are often used interchangeably in the literature on networks, such as in Katz/Shapiro (1994), p. 94. This study also uses the terms interchangeably, but it should be noted that this approach is not without controversy. Liebowitz/Margolis (1998) object to the comingling of these terms, however, especially as concernsindirect network effects. They differentiatenetwork effects from network externalities according to whether the impact of an additional user on other users is somehow internalised. If network effects are not internalised, the equilibrium network size may be smaller than is efficient. They classify network externalities as ‘a specific kind of network effect in which the equilibrium exhibits unexploited gains from trade regarding network participation’ (Liebowitz/Margolis (1994a), p. 135). 7 Cf. Farrell/Saloner (1986), p. 940; and Besen/Farrell (1994), p. 118. 197 Exploring theoretical basics – scale effects in clearing increasing network size, the utility that a user derives from consumption increases over-proportionately. A general distinction can be made between direct and indirect network effects, which can be either positive or negative. 8 Positive direct network effects refer to the benefits a user derives from the consumption of a product that is used by others. The value of some products or services also depends on whether they are offered in isolation or in combination with others. 9 Indirect network effects refer to (positive) externalities that do not result from a direct interconnection with others, but rather through the distribution of complementar y goods. 10 Note that networks can also exhibit negative externalities, which result in costs to users from ‘[c]hanges in the size of an associated network’. 11 A growing interest in investigating the relevance of network theory for analysing the organisation of financial markets has recently emerged in the economic literature. In the following, the concept of networks is applied to clearing. 6.1.1 Network effects Networks are common in financial services. 12 Clearing has very strong network externalities, where the value to a user is greatly increased by the access that is given to a wide range of trading counterparties. 13 Network theory is applicable to a variety of financial services and to most parts of the transaction value chain. When interaction among consumers is important, markets are likely to exhibit strong network effects. 14 Existing network studies primarily apply to the trading function and exchanges; 15 8 Cf. Katz/Shapiro (1985), p. 424; Tirole (1988), p. 405; and Economides (1996), p. 679. 9 Cf. Katz/Shapiro (1994), p. 93. 10 Indirect network effects resulting from a growing number of network users include an increase in the number of complementary goods offered, learning effects from employing new technologies and less uncertainty about the sustainability of a new technology. Cf. Thum (1995), pp. 8–12. Economides/Salop (1992) provide one oftheearliest frameworksand insights into the economics of indirect network effects; studies by Church/Gandal/Krause (2002) show that indirect network effects can give rise to adoption externalities. 11 Liebowitz/Margolis (1994a), p. 134. 12 Economides (1993), Abstract. 13 LCH.Clearnet (ed.) (2006b) p. 5. 14 Cf. Liebowitz (2002), p. 20. 15 See, e.g. the contributions of Economides/Siow (1988); Economides (1993); Domowitz (1995); Economides (1996); Domowitz/Steil (1999); Geiger (2000); Book (2001); Di Noia (2001); Claessens et al. (2003); Hasan/Schmiedel (2004b); Hasan/Schmiedel (2004a); Hasan/Schmiedel (2006); and Hasan/Hasenpusch/Schmiedel (2007). 198 Clearing Services for Global Markets in contrast, little research has been done on post-trading networks. 16 Anew stream of recent contributions finds that securities settlement and safekeeping institutions exhibit features of so-called ‘two-sided platforms’. 17 Besides that, increasing attention is paid to applying the network economic concepts of ‘switching costs’ and ‘standard setting’ to the settlement and safekeeping industry. 18 Although the clearing function is widely considered to exhibit network effects, 19 the author is not aware of any studies that provide a classifica- tion or analysis of clearing-related network effects. Additionally, while some characteristics of ‘classic’ network industries are applicable to the clearing function, the Value Provision Network has several distinctive features that distinguish it from other network industr ies; this makes it difficult to apply standard network economic analyses directly. 20 This chapter provides an exploratory attempt to deliver insight into this issue. To accomplish this, it is necessary to: r elucidate the general idea and basic formation of networks; r apply the network view to the Value Provision Network; and r identify and analyse network effects within the VPN (sections 6.1.1.1– 6.1.1.4). Generally, networks emerge from links that connect complementary nodes. 21 The structure of a telephone network illustrates the basic organisational prin- ciples. Members of this network are connected via a central node (the ‘switch node’, symbolised by S) that enables all members of the network to intercon- nect. A phone call from Customer Node A to Customer Node B consists of 16 The studies of Milne (2002); Holthausen/Tapking (2004);Kauko (2005); andVan Cayseele/Wuyts (2005) are among the few contributions to thisfield. Knieps (2006) provides for a network economic analysis of competition in securities post-trade markets. Although Knieps defines the post-trade functions covered by his analysis to include clearing and settlement, CCP services are left aside. He focuses on the network characteristics of basic clearing services – such as provided by CSDs, ICSDs, custodians or banks. 17 See Kauko (2002); Rochet (2005); Kauko (2005); Van Cayseele/Voor de Mededinging (2005); and Van Cayseele/Wuyts (2005). In two-sided markets, two or more platforms are needed simultaneously to complete a transaction successfully. Refer to Rochet/Tirole (2001); Parker/Van Alstyne (2005); Arm- strong (2006);Rochet/Tirole (2006);and Van Cayseele/Reynaerts (2007) formore details andan analysis of two-sided markets. 18 See Milne (2005); and Serifsoy/Weiß (2005). 19 See, e.g. European Central Bank (ed.) (2001b), p. 82; London Stock Exchange (ed.) (2002), p. 5; Russo (2002), p. 237; LCH.Clearnet (ed.) (2003b), p. 3; Heckinger/Lee/McPartland (2003), p. 9; Singapore Exchange (ed.) (2004), p. 8;BNP Paribas Securities Services (ed.)(2005), p. 3; Office of Fair Trading (ed.) (2005), p. 4; Serifsoy/Weiß (2005), p. 8; Corporation of London (ed.) (2005), p. 56; Van Cayseele/Wuyts (2005), p. 3; Schmiedel/Sch ¨ onenberger (2005), p. 35; Branch/Griffiths (2005), p. 3; LCH.Clearnet (ed.) (2006b), p. 5; LIBA (ed.) (2006), p. 6; Bliss/Papathanassiou (2006), p. 24; and Milne (2007), p. 2945. 20 Cf. Milne (2007), p. 2947. 21 Cf. Economides (1996), p. 674. 199 Exploring theoretical basics – scale effects in clearing S A E C G B H D F A H – CUSTOMER NODES SWITCH NODE S Figure 6.2 Telephone network as a simple star network Source: Economides (1996), p. 675. two connections called AS and BS (see Figure 6.2), which represent comple- mentary components. In the case of the telephone network, all components (AS, BS, etc.) are complementary to each other. 22 This kind of network exhibits positive network effects: when a network with n customer nodes is enlarged by one additional customer node (n+1), 2n new ways for interconnection result. 23 An enlargement of the network thus benefits all members of the network. ‘In a typical network, the addition of a new customer (or network node) increases the willingness to pay for the network services by all participants.’ 24 Networks can also be classified as horizontal or vertical. Whereas in hori- zontal networks (such as the telephone network), members are interconnected to build a network, vertical networks join together complementary goods. 25 Each good is useless in isolation, as the demand for one good is dependent on the demand for a complementary good. 26 Network structure varies according to the characteristics of the relevant industry. 22 Economides/White (1994) differentiate networks in which all components are complementary to each other , which are referred to as ‘two-way networks’, and networks in which only some components are complementary to each other, so-called ‘one-way networks’. Cf. Economides/White (1994), pp. 1–5. A more general distinction differentiates one-way networks as those in which the sensible transactions can flow in only one direction, whereas the opposite is true in two-way networks. 23 Cf. Economides (1996), p. 679. 24 Economides (1993), p. 89. 25 Cf. Gr ¨ ohn (1999), p. 25. 26 Examples of vertical networks include personal computers, operating systems and application software or video-cassette recorders and video tapes. 200 Clearing Services for Global Markets When the network view is applied to the Value Provision Network,a two-layered and two-level structure results (see Figure 6.3). 27 The first level corresponds to the network structure constituted by a CCP, which acts as the central ‘switch node’ by becoming the buyer to every seller and the seller to every buyer, and its clearing members. 28 The second level refers to the network structure established by the clearing members acting as GCMs and their respective non-clearing members. Additionally, the first and second network levels each consist of two lay- ers. A central counterpart y constitutes a two-layered network composed of a system layer and a product layer. All of the CCP’s clearing members are interconnected through a shared clearing system (system layer). The resulting network corresponds to the physical network, which is the electronic clear- ing platform provided by the CCP. The clearing system provides the actual clearing services, which include data transfer and processing, bundled net- work services (e.g. netting and cross-margining) 29 and the guarantee function. Within the clearing system, a subset of clearing members forms horizontal networks in different products. This function is enabled through the net- work’s product layer, 30 which represents the open interest held by a CCP in the respective products. As an example: the network of product A consists of clearing members CM1, CM2, CM6, CM7 and CM8, who demand clearing services for the product, and is represented by the open interest in product A. The networks of (non-fungible) products A and B are not compatible; clearing members cannot close-out open positions in one product by enter- ing into offsetting positions in another product. The two-layered network of the CCP consequently results in a combination of horizontal and vertical networks. 31 27 The basic idea and set-up of the two-layered structure builds on the principles identified by Book (2001) for derivatives exchanges. Cf. Book (2001), pp. 171–9. The idea of additionally classifying clearing networks as two-level structures is the author’s own. 28 The operations of CCPs constitute a unique combination of one-way network activity (position man- agement) with two-way networks. The latter function involves linking counterparties; this means that, essentially, any clearing member of a CCP can act in two ways, i.e. either as a buyer or a seller to the CCP. The network thus works in both ways. Van Cayseele classifies CSDs/ICSDs in a similar way, i.e. as a unique combination of one-way and two-way activities. Cf. Centre for European Policy Studies (ed.) (2004), p. 1. 29 Knieps therefore classifies securities clearing services as value-added telecommunications services. Cf. Knieps (2006), p. 54. 30 A telephone network differs from a clearing network in that the essential relationship between the components is complementary. Within a clearing network, on the other hand, products must possess identical terms in order to be fungible. 31 Note the differences in terminology: classifying a CCP as a combination of vertical and horizontal networks in the context of network economic terminology has to be differentiated from the use of 201 Exploring theoretical basics – scale effects in clearing CCP CM1 CM2 CM5 CM6 CM7 CM8 A CM 3 A B B CM4 System-Layer Product-Layer CM Clearing Members NCM Non-Clearing Members Open Interest Product AA Open Interest Product B B 2ND LEVEL LEVEL OF DIRECT ACCESS TO CCP LEVEL OF INDIRECT ACCESS TO CCP 1ST LEVEL NETWORK LEVELS IN THE VALUE PROVISION NETWORK Clearing House (CCP) NCM1 NCM3 NCM4 NCM5 NCM2 Customers RESEARCH FOCUS Figure 6.3 Value Provision Network as a two-layered and two-level network Source: Author’s own; idea and structure based on Book (2001), p. 171. 32 A clearing member acting as a GCM also constitutes a two-layered network that contains a system layer and a product layer. All of the GCM’s non-clearing members are interconnected through the GCM’s electronic clearing system, which forms the basis for the clearing services provided by the clearing mem- ber. Whereas the system layers of the CCP and the GCM require compatibility, they are usually not, and do not have to be, identical. The clearing member will either employ a proprietary or a vendor solution to ensure the compatibility of its in-house system with the CCP system. Whereas the product layer of the CCP network corresponds to the open interest held by the CCP, the product layer of the GCM network merely mirrors the open positions held by the CCP. The networks of (non-fungible) the term ‘vertically integrated clearing house’, which refers to the integration of various parts of the transaction chain (such as trading, clearing and settlement). 32 Book (2001) classifies derivatives exchanges as a two-layered network, consisting of a product layer and asystemlayer. 202 Clearing Services for Global Markets products A and B are not compatible on the GCM level either; open positions in one product cannot be closed out by entering into offsetting positions in another product. The two-layered network of the GCM is consequently also a combination of horizontal and vertical networks. To summarise, the clearing services offered by CCPs and GCMs are network goods. The value-added of these services is impacted by the number of partic- ipants in the networks. The following provides a classification of the positive and negative network effects on the first (CCP level) and second (GCM level) network levels that result from changes in the number of participants (sections 6.1.1.1 and 6.1.1.2). In each case, the network effects on the system layer and the network effects on the product layer are differentiated. The way and extent to which network effects on the first and second levels impact one another are also examined (section 6.1.1.3). Additionally, the spill-over of the first and second level network effects on to other parts of the transaction value chain – such as trading and settlement – is investigated (section 6.1.1.4). 6.1.1.1 First level (CCP level) network effects To begin with, network effects on the first level of the VPN are analysed (see Figure 6.4). The product layer is subject to four different positive network effects: netting, size, cross-margining and open interest effects. A negative effect, in the form of a systemic risk effect, can also emerge. Positive net- work effects on the system layer include the collateral management, interface, complementary offering, learning and infor mation effects. Negative network effects on the system layer can eventually arise in the form of a performance effect. Finally, the negative monopolistic behaviour effect can arise both on the product and system layers. 6.1.1.1.1 First level networ k effects: product layer An important and direct network effect on the product layer resulting from an increasing number of clearing members is referred to as the netting effect. Multilateral netting facilities, such as CCPs, strongly economise on the total number of transactions. 33 The more clearers are connected to a CCP, the more transactions can consequently be processed v ia the clearing house. 34 This increases the utility to each clearing member, as more transactions are available for netting. This accretion of utility results from lower transac- tion costs throug h enhanced possibilities for netting. The more transactions 33 Cf. Hills et al. (1999), p. 132; Hardy (2004), p. 58; Singapore Exchange (ed.) (2004), p. 8; and Branch/Griffiths (2005), p. 5. 34 For limitations of this statement, refer to section 6.1.1.3. 203 Exploring theoretical basics – scale effects in clearing System LayerProduct Layer Positive Network Effects Negative Network Effects Netting Effect Enhanced netting opportunities Open Interest Effect Certainty regarding sustainability of CCP Systemic Risk Effect Concentration of risk Size Effect Clearing services for more products Cross-Margining Effect Increased opportunities for cross-margining Collateral Management Effect Centralised collateral management Performance Effect Overloading of system capacity Complementary Offering Effect Increased number of complementary clearing services Information Effect Reduced information asymmetry Learning Effect Greater know-how and expertise 1ST LEVEL IN VALUE PROVISION NETWORK Interface Effect Connection to various platforms Monopolistic Behaviour Effect Anti-competitive behaviour Figure 6.4 First level (CCP level) network effects on the product and system layers Source: Author’s own; idea and structure based on Book (2001), p. 173. with the same underlying and the same attributes that are available for pro- cessing through the CCP, the more netting can occur. Risk management costs can thus be reduced. In addition, the more transactions that are available for the netting of payments across multiple contracts, the less collateral the clear- ing member has to deposit at the clearing house. In net margining regimes, clearing members profit from reduced capital costs as well as from reduced risk management costs, as less risk monitoring is necessary. The utility to each clearing member is additionally increased through enhanced settlement efficiency. 35 The greater the number of transactions available for the netting of delivery instructions for cash or securities deliveries or the delivery of another underlying, the lower the number of obligations to be settled will be. This in turn results in lower transaction costs through minimising the fees charged by the intermediaries involved in the settlement process as well as in 35 In the clearing of securities, the so-called ‘settlement efficiency’ is an important indicator for efficiency gains. It is measured by relating the number of executed trades to the number of trades settled. Cf. Devriese/Mitchell (2005), p. 20. Central counterparty clearing commonly results in a settlement efficiency of around 95 to 96 per cent in the cash equity markets and 70 per cent or more in the fixed income markets. Cf. SWX Swiss Exchange (ed.) (2007), p. 10; and LCH.Clearnet (ed.) (2003a), p. 31. 204 Clearing Services for Global Markets 0 20 40 60 80 020406080100120140 No. of Clearing Members in 2005 No. of Cleared Products Figure 6.5 Number of cleared derivatives products and clearing members in 2005 (N=6) 36 Source: Author’s own; based on FOW (ed.) (2001); FOW (ed.) (2003); FOW (ed.) (2006); and clearing houses’ websites. reduced back-office costs resulting from the decrease in back-office errors and back-office handling. The sizeeffect relates to the number of products cleared through a CCP. The higher the number of clearing members that route their transactions through a specific clearing network, the higher the number of cleared products will likely be. Clearing members of a clearing house with a large network are then able to clear more products via the respective CCP than members of a smaller clearing house. This positively impacts transaction costs by allowing clearing members to le verage their in-house IT systems and back-office. The size effect is illustrated by Figure 6.5. 37 The size effect assumes that the number of cleared products is indirectly determined by the number of clearing members connected to a CCP, as these connections represent potential demand for the clearing of additional products. This relationship constitutes an indirect network effect. The number of cleared products then increases with an accretive network size. In the event that the size effect leads to a provision of central counterparty clearing services 36 Equity options and single stock futures are each considered one product and are not counted on a per-stock basis. 37 Nonetheless, due to the small sample selection of clearing houses, further research is needed to provide more convincing evidence. [...]... whether financial intermediaries should engage both in balance and off-balance activities Malkam¨ ki a (1999) and Schmiedel/Malkam¨ ki/Tarkka (2002) estimate translog cost functions for their studies, a relying on a broader sample of data available for their focus area Book (2001) on the other hand copes with the problems related to the limited availability of data for his research by providing an analysis... selection of the world’s major derivatives clearing houses is analysed.112 The basis for the examination is the data for the years 1999 to 2005 Figure 6.8 shows the development of unit costs and the number of contracts cleared for selected clearing houses.113 111 112 113 The sample selection of clearing houses was made based on the availability of data Only a few clearing houses worldwide publish annual reports... cross-margining, open interest, collateral management and information effects can positively impact the liquidity and allocation of risk and capital – thus increasing the efficiency of capital markets. 62 Additionally, the information effect positively influences stability by minimising the impact that disturbances arising from the default of a market participant can have on the safe and sound functioning of. .. clearing houses by correspondence The total cost variable in this study represents the reported operating expenses of a clearing house, including depreciation The variables are based on publicly available information, which can be found 226 Clearing Services for Global Markets The analysis supports that decreasing average costs are related to an increase in the number of contracts cleared The linear... available for quantifying the output of derivatives clearing houses; their adequacy for an empirical analysis is discussed in the following The output of derivatives clearing houses can be quantified through the application of various indicators for its activity of providing clearing services The most important indicators are the number of contracts cleared90 and the corresponding value,91 which can... based on the total costs of clearing, whereas the additional output realised by asset classes other than derivatives is not taken into account.114 A substantiated empirical analysis, i.e through the estimation of a cost or production function,115 cannot be properly conducted due to the limitations of the data and the small sample selection Taking these drawbacks into account, the empirical examination... opportunities and a higher number of transactions available for cross-margining Clearing members benefit from collateral management optimisation through savings in cost of capital and back -of ce costs Additionally, as more collateral is held and managed at a ‘secured place’, i.e with a trustworthy and sustainable CCP, the higher the potential savings in risk management costs Because a less complex infrastructure... complementary clearing services These can include automated brokerage solutions for give-up/take-up transactions, enhanced complementary software and technology, and other services outlined in section 2.1.2.3 The supply of a greater number of complementary clearing services is beneficial for clearing members, as it allows them to use the available CCP network for functions and services that might otherwise have... biased due to the fact that a large number of the clearing houses included in the examination provide not only derivatives clearing, but also securities clearing services and do not report the data separately Furthermore, some of the clearing houses analysed are not independent firms, but part of vertically integrated structures This fact also distorts the research results, because the calculation of. .. multiplication of capacity and does not allow for cost savings Cf Gaughan (2002), p 116 Cf Robinson (1948), p 334; Baumol/Panzar/Willig (1982), p 73; and Junius (1997), p 3 224 Clearing Services for Global Markets in terms of the ‘advantages of specialisation’ obtained through the division of labour.102 In addition, economies of scale can be elucidated mathematically.103 Learning effects can also generate . risks. 212 Clearing Services for Global Markets the clearing of additional products and markets, a clearer has an incentive to connect to an increasing number of interfaces with an accretive network. theoretically enable the internali- sation of the size, interface, collateral management and fungibility effects. Enlarging the range of complementary clearing services offered facilitates the internalisation. cross-margining. Clearing mem- bers benefit from collateral management optimisation through savings in cost of capital and back -of ce costs. Additionally, as more collateral is held and managed at a