Introduction
The demutualization of securities exchanges is a clear trend around the world. Yet whereas awareness of this trend among market participants and regulators has grown rapidly in recent years, the causes of the phenomenon are not widely understood. Indeed, there is even considerable confusion regarding what demutualization actually is, with regulators and scholars unhelpfully conflating the concept with for-profit legal status.[1] This article attempts to clarify the meaning and the drivers of demutualization, with an emphasis on the critical role of advances in trading and communications technology.
Mutuality and floor trading
The traditional model of an exchange as a locally organised mutual association is a remnant of the era before trading system automation. As trading required visual and verbal interaction, exchanges were necessarily designated physical locations where traders would meet at fixed times. Access to the exchange had to be rationed to prevent overcrowding and, when single-price periodic call auctions were prevalent, to ensure that simultaneous full participation was physically feasible.
As trading 'systems' were simply rules governing the conduct of transactions, exchanges were naturally run by the traders themselves as co-operatives. Organising trading floors as companies selling transaction services would have been infeasible, as there was no 'system' distinct from the traders themselves -- only an empty room. Rationing access to the exchange was generally done through a combination of substantial initial and annual membership fees, in order to ensure self-selection by high-volume users. Non-members naturally wished to benefit from the network externalities of concentrated trading activity (commonly referred to as 'liquidity'), and therefore paid members to represent their buy and sell orders on the exchange floor. This is how exchange members came to be intermediaries ('brokers') for investor transactions.
Mutuality and automated trading
The economics of automated auction trading are radically different. The placement and matching of buy and sell orders can now be done on computer systems, access to which is inherently constrained neither by the location nor the numbers of desired access points. In a fully competitive 'market for electronic markets,' the traditional concept of membership becomes economically untenable. As the marginal cost of adding a new member to a trading network declines towards zero, it becomes infeasible for an exchange to impose a fixed access cost, or 'membership fee.' Rather, only transaction-based (i.e. variable cost) charging is sustainable. Indeed, we see this trend clearly among automated exchanges faced with significant competition: Deutsche Börse and OM Stockholm, for example, have eliminated membership fees. The transactors on electronic networks, therefore, come to resemble what are normally considered 'clients' or 'customers' of a firm rather than 'members' of an association. And since an electronic auction system is a valuable proprietary product, not replicable without cost by traders, it is feasible for the owner to operate it, and sell access to it, as a normal for-profit commercial enterprise. This contrasts with a traditional exchange floor, whose value derives wholly from the physical presence of traders.
The fact that an automated exchange can be operated as a commercial enterprise, unlike a traditional floor-based exchange, does not in itself make an economic case for a corporate rather than mutual governance structure. However, such a case emerges naturally from an analysis of the incentive structures under which a mutualized and corporate exchange operate.
Exchange members are the conduits to the trading system, and they thereby derive profits from intermediating non-member transactions. They can therefore be expected to resist both technological and institutional innovations which serve to reduce demand for their intermediation services, even where such innovations would increase the economic value of the exchange itself. If the members are actually owners of the exchange, they will logically exercise their powers to block disintermediation where the resulting decline in brokerage profits would not at least be offset by their share in the increase in exchange value.
As a major economic benefit of automated auction trading is the elimination of the need for trade intermediation, mutualized exchanges can be expected both to have difficulties introducing automated systems and, once introduced, allowing their full potential to be exploited by non-member investors. Both of these effects can be well documented. The largest UK-based market-makers on the London Stock Exchange (LSE) fought to block the adoption of electronic auction trading in the mid-1990s. New York Stock Exchange (NYSE) specialist firms have long fought against automated matching of investor orders and display of their limit order books to the trading floor and the wider public. Nasdaq market-makers blocked the incorporation of mandatory price-time priority in Nasdaq's trading system upgrade, known as SuperMontage, successfully arguing that customers should be allowed to trade with them even if they were not posting the best price, or the earliest best price (they could simply agree to match the best posted price). Non-member-based commercial trading system operators, on the other hand, have always chosen both to operate automated auction structures and to do so without any intermediation requirement (except for retail orders, for which such operators have traditionally not wished to manage the credit risk function).
What are the efficiency costs inherent in exchanges continuing to operate as broker-dealer co-operatives? Recent empirical evidence suggests two primary ones: higher trading costs (and therefore lower returns) to investors, and higher capital costs to listed companies.
Domowitz and Steil (2002) estimate total trading costs to be 28-33% higher through NYSE and Nasdaq traditional broker members than through non-intermediated for-profit trading system operators (now commonly referred to as ECNs). They estimate that European trading fees alone would fall a massive 70% if the European exchanges were to move to an ECN governance model: i.e. eliminating membership and allowing direct investor access. Historically, mutualized exchanges have sought to fix commissions and prevent price competition.[2] For-profit non-member-based trading system operators, on the other hand, have the opposite incentive: to mitigate access costs to their system imposed by intermediaries.
The emergence of commercial rather than mutualized trading operations should also result in lower capital costs to listed companies. Domowitz and Steil (2002) demonstrate not only that distintermediating trading reduces trading costs, but also that trading cost reductions in turn reduce the cost of raising equity capital. The halving of total trading costs which they document in the US between 1996 and 1998 resulted in an 8% decline in equity capital costs to S&P 500 companies. The authors further estimate that the elimination of mandatory broker intermediation at the European exchanges would result in at least a 7.8% saving to European blue-chip companies.
Demutualization
Remarkably, there is no standard definition of a 'demutualized' exchange. In informal discussion, the emphasis is often placed on whether the exchange is run on a for-profit basis. However, the central question is not whether the exchange is legally able to distribute surplus funds back to the owners -- the definition of for-profit status -- but rather who owns the exchange in the first place. For-profit exchanges are therefore not necessarily demutualized; in fact, most are not.
Separation of ownership and membership is fundamental to the concept of demutualization. Why do exchanges choose to bring in non-members as owners? Exchange officials often maintain publicly that they must sell ownership stakes to outsiders as a means of raising capital for expansion and technology investment. Empirically, however, we find that raising capital is generally a secondary aim, or absent as an aim altogether. Most exchanges that have demutualized have had no immediate need for fresh capital. Amsterdam actually used its demutualization as an opportunity to return excess capital to the members. Furthermore, if capital is, in fact, necessary, it can normally be raised from the member firms without having to turn to outsiders.
The primary function of a demutualization is to reduce the control of (particularly local) intermediaries over the strategic positioning of the exchange. This is in recognition of the fact that exchanges operating in a competitive financial market must ultimately be able to reduce capital costs for a significant subset of companies, and raise investment returns for a significant subset of savers, relative to the next-best financing alternative (whether that be another stock exchange, the bond market, or the banking sector). Intermediaries, in seeking to maximise their own profits from trade intermediation, can act to impair the ability of exchanges to serve companies and investors with maximum efficiency. It is not surprising that officials of mutualized exchanges rarely argue this publicly, and instead claim that demutualization is needed to expand their capital bases. But in reality it is typically a response to members frustrating their efforts to implement less intermediated trading structures, to expand direct trading access to foreigners or institutional investors, or to merge with other exchanges. All of these efforts can serve to reduce demand for the services of existing members.
Under what conditions will exchange members actually accept new outside ownership? The key variables tend to be the degree of competition, or potential competition, which the exchange faces, and the degree to which the largest member firms operate internationally.
Competition makes it difficult for members to protect their intermediation franchise, and therefore makes them more open to governance reform and outside ownership. It is not surprising that the pioneer demutualizers were three Nordic exchanges and Amsterdam. These exchanges operate in small and highly open national economies. Each faced significant competitive threats from abroad, particularly London, to trading in their key blue chip stocks. The New York Stock Exchange, on the other hand, has yet to see effective competition materialise, and the members have therefore successfully resisted the high profile demutualization initiative launched by its chairman in 1999.
The internationalization of membership also facilitates demutualization. Large international banks which are members of numerous exchanges have much less motivation to defend mutualization than local players. Locals have a strong incentive to maintain institutional barriers to disintermediation of their services, whereas larger international players tend to see governance reform as an effective weapon for increasing their strategic control of the exchange vis-à-vis the locals (typically by replacing 'one member, one vote' and committee-based decision making with decision-making tied more directly to the size of the ownership stake). The same argument holds for trading automation. Locals tend to dominate market-making and specialist functions. It is no surprise that it was the large international banks which championed the cessation of floor trading in Amsterdam and market making in London in the mid-1990s, whereas the locals fought bitterly to stave off automated trading.
The mere fact that an exchange can be partially owned by specific non-members does not itself suggest that the incentive structure guiding its behaviour will be materially different from that of a wholly mutualized exchange, or that its interests will be better aligned with those of investors and issuers. In many cases, outside ownership is limited to not-for-profit or mutualized entities such as the national central bank or the broker-owned central securities depository (CSD); entities which have little or no incentive to challenge broker interests. Deutsche Börse and the Paris Bourse, prior to their IPOs in 2001, were already organised as for-profit companies, although ownership stakes, controlled almost exclusively by members, could not be sold without approval of the supervisory board (which effectively meant that they could not be sold). These exchanges therefore had an incentive structure almost precisely identical to that of a mutualized exchange. What is essential to a successful demutualization is that non-members are free to buy equity stakes in the exchange from current owners. This is what makes it possible to change the incentive structure.
If we classify as demutualized all exchanges which permit members freely to sell their equity stakes in the exchange to non-members (albeit perhaps with limitations related to maximum shareholdings, etc.), it is still clear that there are huge differences in governance structure among demutualized exchanges. At one extreme is the Borsa Italiana, which is demutualized according to this definition, but is nonetheless 90% owned by Italian intermediaries. By the reckoning of many within the Borsa, it functions more or less as it did prior to its demutualization -- or, more accurately, prior to its privatisation, as it was effectively sold by the Italian Treasury. At the other extreme is OM Stockholm, which is self-listed and which has a highly diversified shareholder base, approximately a quarter of which is foreign. London-based virt-x, a pan-European blue chip exchange dominant only in Swiss SMI index stocks, has a complex ownership structure: 40% of the company is owned by the Swiss Exchange, itself currently a mutual association; 38% is owned by a consortium of major US and European financial institutions; and 22% free-floats on the AIM small cap market of the London Stock Exchange, a competitor.
Table 1 lists those exchanges which we would consider to be demutualized at the end of 2001, and those which may meet this standard in 2002.
Table 1: Exchange demutualizations
| Exchange | Year of demutualization |
|---|---|
| Stockholm Stock Exchange | 1993 |
| Tradepoint / virt-x (never mutualized) | 1995 |
| Helsinki Stock Exchange | 1995 |
| Copenhagen Stock Exchange | 1996 |
| Amsterdam Exchanges | 1997 |
| Borsa Italiana | 1997 |
| Australian Stock Exchange | 1998 |
| Iceland Stock Exchange | 1999 |
| Athens Stock Exchange | 1999 |
| Stock Exchange of Singapore | 1999 |
| SIMEX | 1999 |
| LIFFE | 1999 |
| Toronto Stock Exchange | 2000 |
| Sydney Futures Exchange | 2000 |
| New York Mercantile Exchange | 2000 |
| Hong Kong Stock Exchange | 2000 |
| London Stock Exchange | 2000 |
| Deutsche Börse | 2001 |
| Oslo Exchanges | 2001 |
| Euronext | 2001 |
| Agreements or board proposals for demutualizations and public offerings |
|---|
| Chicago Board Options Exchange |
| Chicago Board of Trade |
| Chicago Mercantile Exchange |
| International Petroleum Exchange |
| London Metal Exchange |
| Nasdaq |
| New Zealand Stock Exchange |
| Nymex |
| Swiss Exchange |
| Tokyo Stock Exchange |
The 1999 (anonymous) FIBV[3] survey found that 11 (22%) of the 49 member stock exchanges responding indicated that they were demutualized, whereas we identified nine (plus two derivatives exchanges and the never-mutualized Tradepoint) at that time. This discrepancy is consistent with an observed general preference among exchanges to be viewed as demutualized. A remarkable 79% of mutualized exchanges surveyed by bta Consulting (2001) claimed that they would demutualize within two years (although the survey does not reveal the number of exchanges polled)[4].
The FIBV survey also found that 49% of respondents indicated that they were organised as limited companies, but not demutualized; 15% were organised as mutual associations; and 13% were effectively state-controlled institutions. 54% of respondents indicated that they were for-profit entities, up from 38% a year earlier. This included 57% of the limited companies and 14% of the mutuals.
Conclusions
Automation of trading structures has been the driving force behind the internationalization of securities trading, inter-exchange competition, and the reform of exchange ownership and governance. Moreover, trading automation is obliterating the traditional competitive boundaries in the industry: between investors and brokers, and brokers and exchanges. As brokers attempt to use technology to internalise more order flow, in competition with the exchanges of which they are members, it is inevitable that the exchanges will sooner rather than later look to outflank their members by marketing their services directly to investors. Indeed, it is only a matter of time before an exchange in some accommodating jurisdiction legally registers itself as a broker, and abandons the membership concept entirely in order to capture unintermediated institutional and retail order flow. Demutualization is only an intermediate stage in the market structure revolution being wrought by trading technology.
References
Banner, Stuart, Anglo American Securities Regulation: Cultural and Political Roots, 1690-1860, Cambridge, UK: Cambridge University Press, 1998.
bta Consulting, Demutualization Survey, February 2001.
FIBV, Cost and Revenue Survey, 1999.
Domowitz, Ian, and Benn Steil, 'Innovation in Equity Trading Systems: the Impact on Trading Costs and the Cost of Equity Capital', in Steil, Benn, David G. Victor, and Richard R. Nelson (eds.), Technological Innovation and Economic Performance, Princeton: Princeton University Press, 2002.
IOSCO Technical Committee, Issues Paper on Exchange Demutualization, June 2001.
Pirrong, Craig, 'A Theory of Financial Exchange Organisation', Journal of Law & Economics XLIII, October 2000.
Notes
[1] See, for example, IOSCO Technical Committee (2001) and Pirrong (2000).
[2] For example, Banner (1998) notes that from the NYSE's founding 'the Board organised brokers into a classic cartel with respect to brokerage commissions' (p.266).
[3] FIBV are the French initials for the International Federation of Stock Exchanges, recently renamed the World Federation of Exchanges.
[4] 78% of exchanges surveyed at the October 2000 FIBV conference said that they either had approval to demutualize or were actively considering demutualization.
This article is based on a larger work entitled 'Changes in the Ownership and Governance of Exchanges: Causes and consequences', published in Brookings-Wharton Papers on Financial Services, 2002.
