Chapter

5 Equity Market Integration

Author(s):
Jörg Decressin, Wim Fonteyne, and Hamid Faruqee
Published Date:
September 2007
Share
  • ShareShare
Show Summary Details
Author(s)

This chapter reviews the progress made in integrating Europe’s equity markets. As noted in Chapter 3, it is more difficult to discern trends in the integration of equity markets than in other markets, in part due to measurement issues. Given this limitation, this chapter addresses the following key questions. How should financial integration in equity markets be measured? Is there something approaching a “European reflex” with respect to investment decisions that leads agents to look beyond national borders? What are the differences among euro area and EU member states with respect to equity market integration? What are the impediments to further and faster integration of national stock markets?

To answer these questions, the chapter develops an analytical framework based on optimal portfolio theory to measure and track equity market integration through the lens of portfolio diversification. Using the concept of “efficient diversification” as a benchmark, the chapter compares performances of EU portfolios, tracks diversification gains over time as a measuring stick of underlying market integration, investigates market and nonmarket factors that may impede a European reflex in investment, and explores policy implications and alternative strategies. But, first, the chapter briefly recounts key market and policy developments affecting EU equity markets.

Recent Market and Policy Developments

Stock market consolidation and regulatory harmonization efforts accelerated after the introduction of the euro.1 In September 2000, the French, Dutch, and Belgian stock exchanges merged to form the Euronext group.2 The horizontal nature of this merger produced a decentralized though technically uniform network structure. Trading rules were largely harmonized and are monitored and enforced by the exchange, but the national markets continue to be regulated by their respective national authorities (Box 5.1). A similar horizontal consolidation process occurred among the Nordic and Baltic stock markets, starting with the 2003 merger into the OMX Group of the Stockholm (OM) and Helsinki (HEX) exchanges.3 In contrast, the Deutsche Börse Group offers an example of a vertically integrated structure, intended to achieve economies of scope by combining trading, clearing, and settlement services in a single institution.4 In horizontal cross-border structures, the “local entry point” determines the relevant set of applicable regulations, and clearing and settlement systems also remain mostly separate, although attempts to integrate them are ongoing (as discussed later).

As described in Chapter 3, the FSAP is the major legislative vehicle for promoting greater financial integration and a more harmonized regulatory landscape within the European Union. The FSAP includes four “Level 1” directives (framework principles) that pertain directly to stock markets:

Box 5.1.The Euronext Model: Preserving Local Market Entry with Technical Unification1

A notable characteristic of the Euronext structure is that local markets were maintained as domestic regulated markets (and as subsidiaries of Euronext NV, the holding company), with the objective of preserving the specific value represented by their local franchise. They represent, for market participants and issuers, “local entry points” from which to access the integrated Euronext market. Although trading rules have been largely harmonized, under the supervision of the exchange, each local market remains subject to its domestic regulation (public law rules) and under the supervision of the local authorities. Hence, for issuers, intermediaries, and investors, the local entry point determines the relevant set of applicable regulations. To better understand the Euronext model, it helps to understand how security trades operate.

The processing of a security order involves three successive phases: (1) the transaction phase, which results in execution of the order; (2) the clearing phase, which allows for the netting of transaction flows in order to reduce the associated risks; and (3) the settlement phase, which results in the effective exchange of securities and cash. Euronext handles only the transaction phase; the clearing and settlement phases are handled by separate but closely associated entities.

The transaction phase. A unified order-driven trading platform is used for cash (NSC) and derivative products (LiffeConnect) across all Euronext markets, under a single set of market rules.

The clearing phase. A central counterparty (CCP)—LCH. Clearnet SA—clears all trades executed on the exchange (the use of a CCP is not mandatory for off-exchange transactions).

The settlement phase. The settlement process is still partially fragmented. Trades concluded on Euronext markets and cleared through LCH. Clearnet SA can settle on the books of members of the Euroclear group (Euroclear bank, Euroclear France, Euroclear Netherlands) or on the books of InterBolsa and CIK, the Portuguese and Belgian central securities depositories (CSDs), respectively, which are both owned by Euronext. Participants may choose to settle their transactions either in the local CSD, which provides access to central bank money, or in Euroclear Bank, the international central securities depository (ICSD), which provides access to commercial money.

1 Prepared by Frantçois Haas.
  • the Market Abuse Directive (Directive 2003/6/EC) aims to prevent insider dealing and market manipulation;
  • the Prospectus Directive (Directive 2003/71/EC) aims to provide firms with a “single passport” to raise investment capital on a pan-European basis;
  • the Transparency Directive (Directive 2004/109/EC) establishes minimum requirements and the mechanisms for periodic financial reporting for issuers whose securities are admitted to trading on a regulated market in the European Union; and
  • the Markets in Financial Instruments Directive (MiFID, Directive 2004/39/EC) aims to provide investment firms with a single passport to operate across the European Union, while ensuring protection for investors.

Although the impact of these FSAP measures will take time to become clear, the key empirical questions are whether and to what extent regulatory harmonization, market-based consolidation efforts, and the introduction of the euro have advanced integration of national stock markets in Europe thus far, and if not, why not?

Measuring Equity Market Integration

Measurement Challenges

Measuring the integration of bond and money markets is often done using convergence or correlation statistics that capture movement toward the law of one price. Extending this approach to stock market returns, Figure 5.1 shows the standard deviation in (filtered) total returns across euro area stock markets.5 The figure shows that the dispersion of returns has clearly decreased since 2000. Abstracting from the 1998–2000 volatility, one might conclude that equity market integration in Europe has substantially increased under European Economic and Monetary Union (EMU).6

Figure 5.1.Dispersion of Euro-Area Stock Market Returns, 1990–2006

(Cross-country standard deviations; in percent)

Source: European Central Bank.

Note: Data through March 2006.

However, these convergence- and correlation-based statistics are imperfect (if not improper) measures of equity market integration and can be misleading, including for the following reasons.

  • Noncomparability of the underlying assets. Whereas fixed-income assets or money market instruments feature a relatively small set of defining characteristics (for example, interest rate, maturity and duration, credit rating), equities represent an ownership stake in a unique company.7 Shareholders purchase an uncertain stream of future cash flows, which depend on the firm’s product line and brand, management team, and business strategy, dividend policies, sectoral and competitive exposures, and so on.8 Furthermore, many of these features can change over time. All this creates differential risk characteristics for each stock.
  • Compositional factors. Even broad stock market indices cannot be presumed to be comparable, given national differences in the structure of production. The relative weights of (say) financial stocks versus technology stocks may look quite different from one equity market to another.9 Moreover, real integration might cause national indices to look more (not less) dissimilar over time.10
  • Common shocks. Increasing correlations between stock markets may also reflect common external shocks, rather than closer equity market integration; if so, rising correlations may be “spurious” as an indicator of increasing integration.11 Since EMU, stock markets in the euro area face common interest rate and liquidity shocks and growing real economic linkages. These developments should increase correlations in stock returns even in the absence of further stock market integration.

In short, correlation- or convergence-type measures are problematic because they are, at best, indirect measures that may be influenced by factors possibly unrelated to equity market integration. Adams and others (2002) aptly summarize the various reservations surrounding returns-based measures: “given the instability of the indicator and the questionable economic interpretation of ex-post return correlations, it appears unwise to draw any conclusions based on such kind of indicators.” Thus, a more coherent framework, placed on a more solid footing with respect to economic theory, is needed to address the relevant measurement and conceptual issues and to track and assess equity market integration. Such a framework is developed below.

Tracking Equity Market Integration

Before describing the analytical framework, it is helpful to briefly review the historical behavior of European stock markets.

Stock Market Data and Descriptive Statistics

Data on total stock market returns (measured in U.S. dollars) for each of the major European equity markets are shown in Figure 5.2.12 The index of total returns reflects both price changes (in other words, capital gains or losses) and dividends paid on an aggregate basis for each of the major stock market indices.13 Data reflect gross returns (not adjusted for taxes or fees), which one should bear in mind later when considering cross-border holdings and comparing actual and “efficient” portfolios. As shown in the figure, EU stock markets have generally moved together, including notably during the boom period of the late 1990s and 2000, and during the ensuing bust and recovery.14

Figure 5.2.European Stock Market Performance, 1990–2005

(Total U.S. dollar returns; January 1999 = 100)

Source: Datastream.

Note: Data for Ireland not available before 1998.

However, around these general trends, there is a good deal of country variation (Table 5.1). A few salient statistical features, evident in the table, are worth highlighting. First, average returns from European equities varied considerably across EU-15 stock markets between 1990 and 2005. On the lower end, total stock market returns for Italy, Portugal, and Germany averaged around 7 percent a year; on the upper end, Greek and Finnish stock markets had annualized returns around 15 percent. Also, the monthly volatility of returns covered a wide range. Greek stocks, for example, fluctuated widely from month to month, typically rising by more than 11 percent or declining by more than 8 percent (± 1 standard deviation around the mean) in about two-thirds of the months; U.K. stocks, by contrast, fluctuated within a relatively narrow range of 3½ percent declines and 5 percent increases on a monthly basis.

Table 5.1.Distribution of Monthly European Stock Market Returns, 1990–2005(Total U.S. dollar returns; in percent unless noted otherwise)
Stock MarketAverage

Return
Standard

Deviation
Skewness1Kurtosis1Normality1ARCH2
Austria0.686.35–0.74**2.02**17.67**18.68**
Belgium0.774.94–0.68**1.99**17.35**0.90
Denmark1.005.02–0.41*0.405.400.02
Finland1.228.71–0.201.24**11.26**0.22
France0.735.53–0.41*0.696.24*5.87*
Germany0.586.46–0.84**2.84**24.22**1.97
Greece1.319.730.90**3.66**31.93**0.25
Ireland1.006.09–0.50*0.904.930.10
Italy0.526.630.020.452.260.54
Netherlands0.915.50–0.85**2.26**20.08**3.84*
Portugal0.555.44–0.180.432.672.45
Spain0.956.29–0.310.77*5.913.00
Sweden0.867.28–0.46**0.72*6.99*8.15**
United Kingdom0.784.34–0.060.050.863.18
Sources: Datastream; and IMF staff estimates.

Test statistics for no skewness, no excess kurtosis, and normality (i.e., Doornik-Hansen test). A * (**) indicates statistical significance at the 5 (1) percent level.

Engel test for first-order autoregressive conditional heteroscedasticity or ARCH(1).

Sources: Datastream; and IMF staff estimates.

Test statistics for no skewness, no excess kurtosis, and normality (i.e., Doornik-Hansen test). A * (**) indicates statistical significance at the 5 (1) percent level.

Engel test for first-order autoregressive conditional heteroscedasticity or ARCH(1).

In terms of higher moments (skewness and kurtosis) of the distribution, several stock markets yielded returns that were negatively skewed (subject to more significant downside corrections) and displayed excess kurtosis or “fat tails” (occasional large realizations in returns). Also, volatility clustering (such as ARCH-type behavior) or intermittent episodes of relatively heightened market turbulence appeared significant in a few markets. Overall, a normal distribution appeared to aptly characterize the behavior of returns for half of these national stock markets, although in some other cases, departures from normality were significant.15 These issues will be pertinent to the analytical framework outlined below, which is based on mean-variance optimization, and to considerations of robustness and sensitivity (discussed later).

Efficient Portfolio Diversification

A fundamental benchmark is needed to assess whether portfolio investment in Europe has become more closely aligned with behavior consistent with an increasingly integrated equity market. In search of a “European reflex” among agents transacting in equity markets, the analysis approaches this issue from the vantage point of efficient portfolio diversification. The analytical framework that is developed here is based on traditional portfolio selection theory as described in Box 5.2. However, the concept of mean-variance efficiency in portfolio allocation is used here in a somewhat novel way—namely, as a yardstick for assessing the state of equity market integration across EU member states.

From the theory of optimal portfolios, one can derive the efficient risk-return frontier for European stocks, where efficiency is defined as the lowest portfolio variance (risk) that delivers a certain rate of return. This frontier represents the best combinations of risk and return that investors holding an EU portfolio could achieve in a given time period.

Compared with the efficient set of diversification possibilities, “distance to the frontier” then represents forgone or potential diversification benefits. More concretely, Figure 5.3 plots the efficient frontier for EU-15 stock returns from 1990 to 2005 based on the solution to the mean-variance optimization problem described in Box 5.2. For illustrative purposes, the figure also shows the historical performance of a uniform portfolio, that is, one that places equal weight on all EU-15 stock markets. This “naïve” diversification strategy would have yielded an average monthly return of about ¾ percent (9 percent annualized) with a monthly standard deviation of 4¾ percent during the sample period. The area or “diversification cone” to the northwest of this interior point, closer to the frontier, reflects superior outcomes that offer either a higher rate of return for the same or lower degree of risk or a lower risk for the same or higher level of return.16

Figure 5.3.EU-15 Efficient Risk-Return Frontier, 1990–2005

(In percent)

Sources: Datastream; and IMF staff estimates.

Underdiversification is the most obvious source of inefficient portfolio allocation and commonly takes the well-known form of “home bias,” the tendency of investors to favor home-country assets in their port-folios.17 Home bias in portfolio allocation decisions may be quite costly, as indicated by the size of the diversification cones for national equity markets returns and risks (Figure 5.4). The figure shows that stock markets in Germany, France, and the United Kingdom have historically underper-formed relative to the efficient frontier (and even, in some cases, against a simple diversified portfolio strategy). The cones indicate that substantial potential welfare gains for German, French, and U.K. investors could be achieved by further diversifying into foreign stocks.

Figure 5.4.EU-15 Efficient Frontier and Major Stock Indices, 1990–2005

(In percent)

Sources: Datastream; and IMF staff estimates.

Box 5.2.The Methodology for Deriving Efficient Portfolios

Following the seminal work of Markowitz (1952, 1959) and Tobin (1958), the mean-variance approach to portfolio selection can be described as the following problem with n risky assets:

where ω is an n-dimensional vector of portfolio weights, V is the variance-covariance matrix of asset returns, and R is a vector of expected returns. The constrained optimization problem essentially minimizes—with respect to individual weights—the portfolio’s overall variance (this is, the quadratic form) for a given portfolio return μ. If short-selling is not permitted, an additional constraint is needed: ωi ≥0, ∀i; see, for example, Jagannathan and Ma (2003).

Taking as given asset prices and returns, the mean-variance approach provides a useful general benchmark for mapping the set of diversification possibilities. A general equilibrium approach to asset price determination (such as the capital asset pricing model or CAPM) requires additional assumptions from the “demand side” (for example, common subjective beliefs regarding the distributions of returns); this usually produces strong results that are at odds with data (such as all agents holding the same “market portfolio” and zero trading in equilibrium).1 A less stringent (albeit more partial) application is to simply map out the best available combinations of risk and return with respect to European investment possibilities and to explore the properties of this efficient frontier. See, for example, Moerman (2004) and Ehling and Ramos (2005).

Some additional (well-known) caveats should, however, be noted. For the mean-variance approach to be meaningful (this implies, among other things, consistency with maximizing expected utility) either utility must be quadratic or the joint distribution of returns must be elliptical (for example, normally distributed). Stock market returns, however, may display aspects of non-ellipticality (such as skewness) and time-varying volatility (for example, ARCH or GARCH properties). Estimation errors may also arise because the true data-generating process for expected returns and their covariances is unknown. Simple measures such as the sample estimator of mean, variance, and covariance can be problematic; see Disatnik and Benninga (2005) and the references cited therein.

1Buiter (2003) recounts that Tobin himself questioned the more rigorous application: “Tobin viewed his model of portfolio choice under uncertainty … as describing the behavior of an individual portfolio manager, and he was not an unqualified admirer of the use made by Sharpe (1964), Lintner (1965) and Mossin (1966) of the mean-variance model. All three authors used the Tobin-Markowitz mean-variance model of portfolio selection as the ‘demand side’ of an equilibrium approach to the determination of asset prices…. (which) means that the common portfolio of risky assets held by each and every portfolio holder has to be the market portfolio of risky assets.”

Unrealized diversification gains, in turn, can be interpreted as an indicator of incomplete financial integration. The reasoning is as follows. Many of the underlying factors that produce home bias can be traced to imperfect integration of national stock markets. Departures from the efficient frontier—derived here on the basis of ex post, gross returns—could reflect some of the following factors affecting investor choice and behavior:

  • incomplete information and informational asymmetries, including national differences in reporting standards;18
  • differences in corporate governance practices or regulations, such as shareholder protections;19
  • structural barriers to cross-border trade, including language differences, information frictions and communication costs, and publicity and awareness (see, for example, Faruqee, Li, and Yan, 2004).
  • differences in taxes (for example, on dividends) on foreign holdings and fees associated with cross-border transactions (such as clearing and settlement costs); and
  • exchange rate fluctuations and other (ex ante) risks.

The last point highlights the fact that the efficient frontier is derived from the behavior of realized returns that are unknown and subject to uncertainty at the time investors make their portfolio decisions (see footnote 16). In the presence of incomplete information, unanticipated shocks, and lack of perfect foresight, agents will be unable to precisely attain the efficient diversification benchmark.

Other (nonrival) explanations for departures from efficient portfolio diversification involve technical and conceptual issues surrounding the mean-variance optimization approach. These include the following:

  • Model risk. The possibility of abnormal returns raises questions about whether risk and return are sufficient conditions for understanding investor decisions. More precisely, normally distributed returns have been shown to be a sufficient condition but this may not be met for some markets.20 The economic validity of the mean-variance approach will then depend on how good the second-order approximation—limited to risk and return—is for capturing expected utility maximization.21
  • Estimation risk. While the true underlying risk and return of assets is unob-servable and must be inferred, ex post or observed returns may not yield an undistorted picture.22 Replacing ex post returns with an unobservable measure, though, would introduce a joint-hypothesis problem.23
  • Feasibility issues. A well-known issue with the mean-variance optimization algorithm is that it can generate “unbalanced” portfolios that could include large negative weights (short positions) on some assets.24 However, an approach imposing (at least approximately) short-selling constraints can also be derived as a robustness check (see Appendix 5.1).

Because of these various behavioral, conceptual, and technical issues, the absolute level or size of the diversification cone is only broadly indicative as a measure of diversification (in)efficiency.25

However, changes in the distance to the frontier or the diversification cone can be interpreted as tracking integration gains over time. The argument has two parts. First, assuming that risk and return are the central variables concerning investors’ payoffs and welfare, moving toward the efficient frontier should be desired by economic agents and pursued by them given their ability to do so. Second, the attenuation—through stock market integration—of market and policy barriers and impediments (which help explain departures from efficiency in the first place) should, in turn, enable moving toward the efficient frontier, other things being equal. As agents can better avail themselves of more efficient portfolio allocations, market integration should thus promote greater diversification and risk sharing as discussed in Chapter 2.26 Many of the directives targeted by the FSAP, for example, aim to tackle such national barriers.

European Portfolios and Calculating Home Bias

To assess the extent and efficiency of EU portfolio diversification, data from the IMF’s Coordinated Portfolio Investment Survey (CPIS) are used to calculate cross-border holdings of equities.27 The survey furnishes a geographical breakdown of portfolio investment holdings of equities and other securities (short- and long-term debt instruments) for the pilot year 1997 and annually since 2001. The stock data cover nonresident holdings for 70 countries—including all EU-15 states.28 Using these data, estimates of portfolio shares allocated to other EU-15 stock markets and the degree of home bias can be computed, subject to the data issues discussed in Appendix 5.1.

Table 5.2 presents the CPIS data for 1997 and 2003. Destination countries are given in the rows, and source countries are given in the columns; entries reflect the country shares of nonresident equity holdings in percent of total (nonresident) holdings from all EU sources, except the last column for non-EU countries, which is expressed in percent of nonresident holdings from all sources. The table shows that ownership of nonresident holdings has shifted toward the European Union—particularly, euro area—investors since 1997.

Table 5.2.Ownership Distribution of Nonresident Equity Portfolio Investments in EU-15 Countries, 1997 and 2003(In percent of total stock of nonresident equity portfolio investments held by residents of EU-15 countries, unless noted otherwise)
FromFrom
AustriaBelgiumDenmarkFinlandFranceGermanyGreeceIrelandItalyLuxembourgNetherlandsPortugalSpainSwedenUnited KingdomOther

countries

(in percent

of total

nonresident

holdings)
Into19972003199720031997200319972003199720031997200319972003199720031997200319972003199720031997200319972003199720031997200319972003
Austria1.91.93.43.80.40.37.13.839.60.01.25.312.57.09.821.84.50.40.00.10.32.70.748.522.960.148.1
Belgium1.80.60.90.40.60.424.231.37.90.00.32.43.92.535.615.08.90.80.91.31.81.50.849.66.648.830.7
Denmark0.30.50.91.41.93.55.74.04.70.12.83.35.21.841.512.08.50.10.00.60.116.37.154.223.566.655.0
Finland0.30.93.52.55.81.75.411.019.10.12.23.91.94.112.07.56.10.00.20.52.531.610.141.225.969.245.1
France0.51.14.911.42.61.20.31.220.30.11.74.98.88.114.710.89.30.20.23.05.23.02.664.119.859.744.6
Germany3.75.13.03.33.01.30.30.910.923.30.12.35.05.47.021.617.57.60.20.43.04.84.92.545.917.055.347.4
Greece0.56.03.70.30.60.49.76.99.11.05.056.98.315.41.512.30.00.01.70.82.21.120.635.841.649.6
Ireland1.01.318.11.60.21.00.01.911.812.113.50.08.625.08.71.05.10.40.10.43.22.51.356.025.257.435.2
Italy0.60.61.12.42.51.00.10.613.822.413.70.01.87.119.66.26.80.10.50.73.72.01.771.219.759.534.4
Luxembourg2.01.540.312.50.00.10.40.711.07.230.70.20.20.023.731.73.31.91.90.75.53.510.75.71.03.68.420.2
Netherlands1.41.213.45.92.71.00.41.119.423.220.00.12.24.65.17.311.20.20.20.92.84.22.150.119.370.448.5
Portugal0.30.61.72.73.41.10.20.814.413.99.50.13.13.74.817.214.85.114.84.119.00.21.862.70.072.044.6
Spain0.10.72.52.71.21.30.10.77.825.417.80.00.94.810.24.912.86.86.30.51.61.21.768.719.142.439.7
Sweden0.60.51.01.610.211.22.617.85.74.410.30.12.04.32.33.315.79.98.30.10.10.50.465.022.170.451.8
United
Kingdom1.51.14.32.55.12.40.72.117.012.112.10.318.915.67.75.615.631.918.20.50.43.94.98.57.182.869.0
Sources: IMF staff calculations, based on IMF Coordinated Portfolio Investment Survey (CPIS) data.Note: Excluding the last column, data in each row add up to 100 for a given year. Numbers for 1997 and 2003 are not comparable. Everything else equal, the 1997 percentages are larger because they are calculated excluding holdings by residents of Germany, Greece, and Luxembourg.
Sources: IMF staff calculations, based on IMF Coordinated Portfolio Investment Survey (CPIS) data.Note: Excluding the last column, data in each row add up to 100 for a given year. Numbers for 1997 and 2003 are not comparable. Everything else equal, the 1997 percentages are larger because they are calculated excluding holdings by residents of Germany, Greece, and Luxembourg.

Two broad trends emerge. First, the relative roles of U.K. investors and, second, non-EU investors have declined significantly in nonresident holdings. Part of this decline is exaggerated by the exclusion of a large EU member state (Germany) and an emerging financial center (Luxembourg) from the 1997 survey, inflating the contributions from other source countries in the initial data. But adjusting for differences in country coverage does not alter the underlying picture:

  • Between 1997 and 2003, the share of nonresident portfolio investment securities owned by non-EU-15 countries declined from 62 to 55 percent of all nonresident holdings (excluding Germany and Luxembourg as source countries in both datasets).
  • More dramatically, the share of nonresident holdings from the United Kingdom declined by more than half, from 39 percent to 17 percent of EU totals (again excluding Germany and Luxembourg). This is despite the important role of London as a financial center.29
  • Among Euronext countries France, Belgium, and the Netherlands, equity cross-holdings have generally increased (Table 5.3). In Belgium, for example, portfolio investment holdings from France as a share of EU-15 totals rose dramatically, from 24 percent in 1997 to 55 percent in 2003 (excluding Germany and Luxembourg). Smaller increases in the share of bilateral holdings were apparent vis-a-vis the Netherlands.
Table 5.3.Change in Equity Cross-holdings between Euronext Countries, 1997 versus 2003(In percentage points)
Source
DestinationBelgiumFranceNetherlands
Belgium31.20.7
France12.63.5
Netherlands–4.814.4
Sources: IMF Coordinated Portfolio Investment Survey (CPIS); and IMF staff estimates.Note: Data represent percentage point changes in nonresident holdings (as a share of total nonresident investments held by EU residents) among the Euronext countries, after adjusting the 2003 data by excluding Germany, Greece, and Luxembourg to ensure comparability.
Sources: IMF Coordinated Portfolio Investment Survey (CPIS); and IMF staff estimates.Note: Data represent percentage point changes in nonresident holdings (as a share of total nonresident investments held by EU residents) among the Euronext countries, after adjusting the 2003 data by excluding Germany, Greece, and Luxembourg to ensure comparability.

Notwithstanding these compositional shifts in equity cross-holdings, home bias remains high in Europe (Table 5.4). Home bias is defined here as the share of resident holdings of domestic stocks in a country’s overall portfolio of EU equities (excluding Luxembourg).30 Using this data-dependent definition, home bias has declined some since 1997, but remains high on average, at around 80 percent. There is a fair amount of country variation, with smaller countries generally exhibiting larger reductions in home bias. Compared with the European Union as a whole, the euro area—which started from a very similar level—saw a marginally larger reduction in home bias.

Table 5.4.“Home Bias” in EU Equity Portfolios

(In percent)1

199720032
Austria86.169.2
Belgium86.970.5
Denmark87.280.9
Finland96.879.0
France93.084.0
Germany
Greece
Italy92.083.7
Netherlands87.773.5
Portugal98.090.5
Spain97.092.4
Sweden88.987.3
United Kingdom90.988.6
EU average91.381.8
weighted by market cap91.385.2
Euro area average92.280.3
weighted by market cap92.082.9
Sources: IMF Coordinated Portfolio Investment Survey (CPIS); and IMF staff estimates.

Domestic equity holdings as share of EU equity holdings excluding Luxembourg as a destination country.

Drops Germany, Greece, and Luxembourg as source countries from 2003 data.

Sources: IMF Coordinated Portfolio Investment Survey (CPIS); and IMF staff estimates.

Domestic equity holdings as share of EU equity holdings excluding Luxembourg as a destination country.

Drops Germany, Greece, and Luxembourg as source countries from 2003 data.

Tracking Integration Gains

Integrating the analysis of efficient risk-return frontiers and the construction of European equity portfolios allows for measures of diversification efficiency and distance to the frontier to be estimated. These are presented for the EU-15 in Table 5.5. In terms of efficient portfolio diversification, Europe displays wide disparities. From a mean-variance efficiency standpoint and relative to the uniform EU portfolio’s performance (normalized = 100 percent), country portfolios range from 6 percent efficiency (Greece) to more than 200 percent efficiency (United Kingdom) with respect to the risk-adjusted returns they delivered for investors in 2003.31Figure 5.5 shows the location of country portfolios vis-à-vis the efficient risk-return frontier.32 In many countries, including the Netherlands and Belgium, better performance reflected greater diversification and smaller home bias, but in others, such as the United Kingdom, it reflected the strong relative performance of local stock markets where holdings were concentrated. Thus, having a large share of local stocks does not imply that one’s portfolio will necessarily underperform. If an investor resides in a home market that delivers superior stock returns at relatively low risk, a high share of domestic stocks would be desirable. This finding highlights a clear shortcoming of the simple (data-based) metric and underscores the need to compare against an efficiency or optimal benchmark in defining “home bias.”

Table 5.5.EU Equity Portfolio Diversification Efficiency(In percent; EU portfolio = 100)
Austria34
Belgium100
Denmark150
Finland16
France43
Germany22
Greece6
Italy22
Netherlands76
Portugal42
Spain30
Sweden15
United Kingdom224
EU average60
weighted by market cap96
Euro area average39
weighted by market cap38
Sources: Datastream; IMF Coordinated Portfolio Investment Survey (CPIS); and IMF staff estimates.Note: Based on 2003 CPIS data; Datastream Total Returns Indices (in U.S. dollars), 1990–2003.
Sources: Datastream; IMF Coordinated Portfolio Investment Survey (CPIS); and IMF staff estimates.Note: Based on 2003 CPIS data; Datastream Total Returns Indices (in U.S. dollars), 1990–2003.

Figure 5.5.EU-15 Efficient Frontier and Country Portfolios

(In percent; EU portfolio = 100)

Sources: Datastream; and IMF staff estimates.

Note: CPIS = Coordinated Portfolio Investment Survey; GMV = global minimum variance.

Based on changes in diversification efficiency, Europe appears to have made modest progress toward more integrated equity markets.33Table 5.6 suggests that the average improvement has been between 10 and 20 percent for the EU-15 and euro area as a whole. Diversification gains have varied at the country level, however. Some countries that had weaker starting points, such as Austria, Finland, and Italy, showed appreciable gains in the performance of their stock portfolios, reflecting more diversified holdings. As a strong performer, the United Kingdom also showed improvement. Interestingly, for Euronext countries other than France, greater country diversification did not benefit portfolio efficiency. For Belgium and the Netherlands, more diversification did not produce “better” diversification, as portfolio reallocation was concentrated in a single country (France). These results suggest that partial integration may be less than ideal if it causes investment diversion of holdings into alliance exchanges.34 What is needed is the ability for companies and investors to list and invest freely, without artificial hindrances, in competing stock markets.

Table 5.6.EU Equity Diversification Efficiency Gains(In percent)
Austria65
Belgium–20
Denmark9
Finland34
France16
Germany
Greece
Italy50
Netherlands–55
Portugal11
Spain24
Sweden16
United Kingdom28
Sources: Datastream; IMF Coordinated Portfolio Investment Survey (CPIS); and IMF staff estimates.Note: Based on 2003 CPIS data; Datastream Total Returns Indices (in U.S. dollars).
Sources: Datastream; IMF Coordinated Portfolio Investment Survey (CPIS); and IMF staff estimates.Note: Based on 2003 CPIS data; Datastream Total Returns Indices (in U.S. dollars).

Impediments to Integration

Despite the regulatory harmonization and market consolidation efforts under way, the multiplicity of European systems continues to create barriers to efficient cross-border trading, clearing, and settlement.35 Market participants must operate under many national systems, each with different technical requirements and practices. These include different settlement periods and operating hours, absence of guaranteed intraday finality of transactions in some markets,36 as well as different tax and legal frameworks. Although many of these differences are being addressed, in part thanks to the work of the Clearing and Settlement Advisory and Monitoring Expert (CESAME) group,37 remaining differences are significant and impose risks and costs on agents. These include the following:

  • For issuers, it remains an open question whether a “European passport” will sufficiently compensate for the absence of a fully integrated European market for equities.
  • For traders, fragmented market infrastructures substantially increase cross-border transaction costs. The related “turnover tax” implicit in these costs, which are four to seven times higher than for domestic transactions or for transactions in the United States (Schmiedel and Schonenberger, 2005), are essentially prohibitive for trading strategies that require even a modest degree of portfolio turnover.
  • Competing trading platforms for dealing in the same securities also tend to reduce scale economies and fragment market liquidity, leading to higher bid-ask spreads. On the other hand, competition between trading platforms reduces monopoly rents.
  • For investors, high fees on cross-border transactions tend to tilt investment toward local alternatives, promoting home bias by reducing net returns from foreign securities. To the extent that these high costs interact with other impediments such as differences in taxes, reporting standards, information, or language, the discouraging effects may be more pervasive. Scale economies from further consolidation in clearing and settlement systems would also lower costs below current levels for all transactions, not just for cross-border transactions (Schmiedel and Schönenberger, 2005; Schmiedel, Malkamäki, and Tarkka 2006).

Interoperability, alliances, mergers, and joint ventures constitute various approaches to market integration. Europe has seen and is seeing examples of each type, and it can be anticipated that this trend will continue. For instance, there have been several partial mergers, resulting in the emergence of Euronext, OMX, Deutsche Börse, and Euroclear, and there also have been alliances and joint ventures in all market infrastructure segments. Interoperability is also increasing, thanks to both private and public sector initiatives. Nevertheless, further integration, including of underlying market infrastructures and processes, will take considerable time, and it remains unclear which approach toward achieving an integrated or consolidated system will eventually dominate and what degree of integration it will deliver.

Policy Lessons

The policymakers’ challenge is how best to conceptualize, facilitate, and steer these market-based consolidation efforts. In light of the many complex impediments to integration, policymakers must make trade-offs to entice and enable agents and markets to pursue further equity market integration. In particular, there is a trade-off between relying on market solutions and on policy initiatives. However, in the absence of public policy initiatives that address legal and other obstacles (such as a number of the Giovannini barriers), the scope for market forces to achieve market integration is very limited.

Cross-listing of stocks on multiple exchanges, for example, provides one avenue for firms to facilitate cross-border equity investments in the absence of integrated market infrastructure. While it facilitates operations between stock exchanges and existing market arrangements, cross-listing does not really integrate underlying market infrastructures or trading practices. It merely creates parallel markets for the same asset, possibly fragmenting liquidity in the process (Box 5.3).

Mergers or alliances between market infrastructures also rely predominantly on the action of market participants. While these can lower some intermarket barriers and costs, even merged providers cannot fully integrate their infrastructures in the current context, as illustrated by the maintenance by Euronext and OMX of national market and infrastructure legs. Moreover, partial mergers or alliances may be problematic if they reduce cross-border trading costs between some EU countries but not between others. As illustrated by the empirical analysis in this chapter, this leads to distortions and investment diversion that distract from full diversification efficiency.

All in all, market-led efforts alone face an uphill struggle to achieve greater consolidation of trading and post-trading infrastructures, given that a number of barriers stand in the way of further consolidation, including legal differences, a lack of standardization, and the vested interests of some agents that benefit from existing arrangements (see Appendix 5.2; see also Schmiedel and Schönenberger, 2005). The limitations of market solutions confer on integration policies an instrumental role in reducing or removing such barriers, widening market access, and better enabling market forces to compete more directly.

Box 5.3.Cross-Listing and European Integration

Cross-listing of shares across national stock exchanges provides one avenue for increasing cross-border trading. For an issuing company, the main potential benefit from listing its stock on multiple exchanges is greater access (at lower cost) to equity finance through a wider investor base. Cross-listing can mitigate market segmentation by reducing barriers to foreign investors that arise from regulations or lack of information (“awareness” hypothesis). Also, meeting local reporting standards can help address information asymmetries, and businesses’ reputations could be enhanced if more stringent financial disclosure or corporate governance requirements need to be met through cross-listing (“signaling” or “bonding” hypothesis). Transaction costs could be reduced for investors (lower bid-ask spread), partly through gains in market liquidity following a foreign listing (Karolyi, 1998).

Multiple listings, however, also entail higher costs. In addition to listing fees, there are often compliance costs related to operating under several sets of rules, and there is a need for arrangements (with attendant costs) for clearing and settlement in the foreign market.

Financial integration and cross-listing have an ambiguous relationship. In more fragmented markets, multiple listings help circumvent trading barriers. As markets integrate, the cost of cross-listing comes down, but so do the incentives and benefits. In Europe, Wójcik, Clark, and Bauer (2004) find that companies cross-listed on other European markets did not gain a corporate governance ratings advantage, but those cross-listed on U.S. exchanges did. Foreign company listings on European exchanges—which typically have higher trading costs and fewer shareholder protections—have generally declined since the 1980s, while European company listings on non-European exchanges (mainly in the United States) have risen. Listing firms tended to be export-oriented and to rely less on debt financing (Pagano, Röell, and Zechner, 2002). Pagano and others (2001) find that European cross-listed companies are also attracted to exchanges that are more liquid and larger, and to markets where other companies from the same industry have already cross-listed.

Cross listings also typically carry higher recurrent costs and compliance burdens for firms than more direct forms of integration, such as merging of European bourses or unifying their underlying trading infrastructures. Moreover, it can fragment liquidity over different markets, with adverse consequences for pricing, volatility, and the depth of the market.

In short, cross-listing provides a cumbersome and unsatisfactory solution for stock market integration in Europe.

In seeking fuller integration, what matters is that all EU investors can access the market or markets offering the highest liquidity at low and equal cost. From the viewpoint of issuers, an integrated market means that they can reach a wider investor base and a deep pool of liquidity with a single, low-cost listing. Interoperability between systems with different rules, regulations, and practices tends to fall short of these objectives. Public monopolies could achieve some of these objectives but tend to lack incentives to remain efficient. So, some degree of convergence, competition, and choice related to listing venues seems advisable.

With MiFID, the European Union clearly opted for competition between alternative market infrastructures, including trading outside regulated exchanges.38 However, competition between marketplaces requires integrated clearing and settlement systems that allow any investor to trade cheaply on any marketplace in the European Union. For now, it is not clear whether integrated clearing and settlement systems can best be achieved by putting in place a public or semipublic monopoly or through competition with improved interlinking between national systems (see Appendix 5.2).

Ultimately, what is required is action both by market participants (whose interests are not homogeneous) and by public authorities, in some desirable combination. While market-led efforts are necessary to achieve a single market for financial services in Europe, complementary, policy-based efforts are also needed to set the stage for more timely and deeper equity market integration. Decreasing the cost of cross-border transactions and reducing differences among market practices and regulations would go a long way toward strengthening a European reflex among investors and firms, and progressing toward a more pan-European capital market. The potential rewards, in terms of fostering risk sharing and portfolio diversification, appear to be significant.

Appendix 5.1. Data, Measurement, Robustness, and Sensitivity Issues

This appendix further discusses issues surrounding data and measurement, the robustness of the estimates surrounding home bias, and diversification efficiency.

CPIS Data and Measuring Home Bias

When considering the CPIS data used to construct country equity portfolios and to estimate home bias, several key measurement issues are worth bearing in mind—some evident from Table 5.2—in the context of examining European portfolio diversification:

  • CPIS data do not provide “own” shareholdings (resident holdings of equities from domestic issuers). These must be inferred residu-ally from data on all nonresident holdings and total stock market capitalization.
  • The CPIS survey may not always provide data on (the correct concept of) a creditor-debtor basis, but instead may provide them on a transactor basis—that is, with geographic attribution based on country of residence of the first party to acquire the security. This may distort measures of underlying holdings, particularly for financial centers such as Luxembourg.
  • CPIS data may exclude certain nonresident holdings if deemed to be of “lasting interest”—that is, direct (rather than portfolio) investment. This could underreport larger holdings in foreign portfolios, overestimating the degree of home bias.39

To handle the financial-center case of Luxembourg, two approaches are possible. First, portfolio holdings attributed to Luxembourg (and Germany) as source countries (significant in many cases) could be dropped from the 2003 data. This would ensure the same country coverage for the 1997 and 2003 data, allowing for comparison. Alternatively, rather than discard this information, Luxembourg data could be distributed to other countries on a “routing center” principle. Namely, equity holdings on an outgoing basis from Luxembourg could be reallocated to other countries according to the geographic pattern of all portfolio investment (equities and debt securities) on an ingoing basis into Luxembourg. The relative implications of the two approaches and the raw data are discussed in the next section.

Table A5.1.Alternative “Home Bias” Estimates in EU Equity Portfolios

(In percent)1

19972003a22003b3
Austria81.762.060.0
Belgium67.953.450.6
Denmark87.280.579.4
Finland96.176.472.3
France91.381.680.3
Germany68.8
Greece97.8
Italy85.465.663.9
Netherlands86.971.968.2
Portugal94.584.784.3
Spain94.889.989.5
Sweden83.280.980.4
United Kingdom90.987.787.2
EU average87.375.975.6
weighted by market cap89.381.378.8
Euro area average87.373.273.6
weighted by market cap88.477.274.5
Sources: IMF Coordinated Portfolio Investment Survey (CPIS); and IMF staff estimates.

Domestic equity holdings as share of EU equity holdings including Luxembourg as a destination country.

Drops Germany, Greece, and Luxembourg, as source countries from 2003 data.

Reallocates Luxembourg 2003 holdings on a “routing center” basis.

Sources: IMF Coordinated Portfolio Investment Survey (CPIS); and IMF staff estimates.

Domestic equity holdings as share of EU equity holdings including Luxembourg as a destination country.

Drops Germany, Greece, and Luxembourg, as source countries from 2003 data.

Reallocates Luxembourg 2003 holdings on a “routing center” basis.

Alternative Home Bias Estimates

The estimates of home bias shown in Table 5.4 reflect the share of resident holdings of domestic stocks in that country’s overall portfolio of EU equities (excluding Luxembourg). Table 5.4 drops CPIS data for Germany, Luxembourg, and Greece for comparability purposes—these countries were missing (as source countries for portfolio investment holdings) from the 1997 dataset. As noted previously, the data for Luxembourg (as a destination country) are further complicated by its role as a financial center, wherein funds may be held as an intermediate point before being channeled to their final destination. Alternative treatments of these countries, however, can be considered to cross-check for sensitivity.

Alternative calculations for home bias in country portfolios are shown in Table A5.1, using several approaches. First, for both the 1997 and 2003 CPIS data, Luxembourg is retained as a destination country for portfolio investment in equities. As shown in columns 1 and 2, this would naturally lower home bias in both years, reflecting the increased holdings of Luxembourg portfolio investment assets and the correspondingly lower share of local assets in other country portfolios. However, the change in home bias from 1997 to 2003 is similar to that reported in Table 5.4, from which these Luxembourg assets are dropped.

Another approach would be to retain these calculations but to further reallocate Luxembourg’s “own” holdings to other countries, based on a “routing center” basis. Namely, Luxembourg’s portfolio equity holdings are allocated to investors in other countries based on the pattern of financial in flows into Luxembourg. This treats Luxembourg as only an intermediate point rather than a final destination for equity portfolio investment.40 Home bias estimates reallocating Luxembourg’s holdings to others is shown in column 3 of Table A5.1, which produces marginal changes (with the possible exception of Belgium).

Note that the measure of home bias in these tables and those in the text are purely data based, devoid of an economic benchmark against which to compare, and it is therefore more aptly labeled “home share.” Model-based estimates of home bias—for example, deviations from the CAPM—can be found in Mann and Meade (2002), De Santis and Gérard (2006), and Baele and others (2006). The last of these also discuss several alternative model-based estimates that all define home bias against a specific set of optimal portfolio weights. This approach tends to suffer from highly sensitive estimates of home bias and the tautology that home bias itself relies on the validity of the benchmark itself. The concept of “diversification efficiency” developed in the chapter provides another benchmark against which “home bias” can be viewed.

Calculating Diversification Efficiency Gains

In Table 5.6, the change in diversification efficiency is shown based on log changes in the area represented by the diversification cone. Percent change formulas yield qualitatively similar but more volatile (outlier) estimates due to small base values in either the initial or end year. Table A5.2 shows the diversification efficiency in each year individually. Note, however, that simply subtracting the entry values for 2003 and 1997 in this table would ignore the (passive) change in the diversification efficiency of the uniform portfolio (that is, the reference benchmark) itself during that time.

Table A5.2.EU Equity Portfolio Diversification Efficiency(In percent; EU portfolio = 100)
19972003
Austria634
Belgium120100
Denmark93150
Finland616
France2343
Germany22
Greece6
Italy522
Netherlands20376
Portugal2442
Spain1330
Sweden815
United Kingdom89224
Sources: Datastream; IMF Coordinated Portfolio Investment Survey (CPIS); and IMF staff estimates.Note: Based on 1997 and 2003 CPIS Data; Datastream Total Returns Indices (in U.S. dollars).
Sources: Datastream; IMF Coordinated Portfolio Investment Survey (CPIS); and IMF staff estimates.Note: Based on 1997 and 2003 CPIS Data; Datastream Total Returns Indices (in U.S. dollars).

Short-Selling Constraints

As discussed in the text, the efficient risk-return frontier raises feasibility issues since part of the frontier may require fairly unbalanced portfolio weights, including large negative weights on some stock markets. While there are economic justifications for why individual optimal portfolios may include large short positions, this appears implausible when considering aggregate or country-level portfolios.41 Imposing short-selling (i.e., inequality) constraints, however, prevents the necessary derivation of an analytical solution.42 To derive the frontier analytically, a local approximation method can be used. Specifically, the dimensionality of the variance-covariance matrix of stock returns can be reduced (by “zeroing out” markets with large negative portfolio positions) sequentially. This method can be shown to produce a (local) solution that preserves non-negativity in portfolio weights in the neighborhood of the GMV portfolio—this is, the region relevant for the calculations of the diversification cone; see Figure A5.1. Thus, diversification efficiency and gains can be (approximately) estimated in the presence of short-selling constraints.

Figure A5.1.EU-15 Efficient Frontier and Short-Selling Constraints

(In percent)

Sources: Datastream; and IMF staff estimates.

Note: GMV = global minimum variance.

Appendix 5.2. Integration of Clearing and Settlement Systems in Europe43

The fragmentation of EU securities clearing and settlement systems impedes the integration not only of securities markets but also of markets that use securities as collateral, such as repo markets. This appendix reviews the main technical and economic features of clearing and settlement systems that make integration necessary but, at the same time, can impede progress toward it. It also discusses policies to achieve integration.

Components of a Securities Transaction

A securities transaction encompasses a value chain with various complementary components. After a securities trade has been executed, a matching process confirms the agreement of the parties on the terms and conditions of the trade. The subsequent clearing process involves determining the obligations of the counterparties, on a gross or net basis. Netting could be offered by either a settlement system (settlement netting) or a CCP, which interposes itself between buyer and seller and assumes their respective rights and obligations. These obligations are then discharged through the settlement process, which ensures delivery of the securities to the buyer and execution of the payment to the seller. Custody includes the issuance of securities either in certificated or dematerialized (book-entry) form, safekeeping of securities on behalf of the customer, and the provision of related services, in particular corporate events such as income payments, redemptions, stock splits, capital increases, proxy voting, and reporting. In most European countries, securities are issued by CSDs, while the value-added services are provided by custodian banks. However, in recent years, competition between these entities has increased markedly due to the deregulation of securities markets.

Characteristics of the Clearing and Settlement Industry

The clearing and settlement industry is characterized by the following:

  • network externalities—the benefits to participants depend on the number of other participants with whom they can do business through a system;
  • economies of scale—high fixed infrastructure costs mean that the average cost per transaction diminishes with the number of transactions;
  • economies of scope—efficiency gains can be achieved from the joint operation of the various components of the clearing and settlement value chain, and these efficiency gains extend to participants, who benefit when they do not have to set up multiple interfaces and implement different procedures for the various components; and
  • high switching costs—incompatibility between systems and the need for relation-specific investments make it costly for participants to switch to another system.

The above features tend to produce natural monopolies. Tellingly, in the United States, post-trade services for equities are provided by a single national user-owned company, the Depository Trust & Clearing Corporation (DTCC). The DTCC emerged from a market-based process triggered by legislation that empowered the Securities and Exchange Commission (SEC) to mandate cost-free interconnectivity. Another monopolist, the Federal Reserve’s Fedwire Securities Service, provides custody and settlement for Treasury bonds and asset-backed securities.

Models for Cross-Border Clearing and Settlement

There are two primary models for cross-border integration of clearing and settlement systems: interlinking national infrastructures across countries and consolidating infrastructures through mergers and alliances.

The interlink model allows the cross-border transfer of securities from one national system to another through technical connections between national clearinghouses and securities settlement systems. A national system acts as a “nominee” on behalf of its participants when securities are transferred to a foreign national system. In addition to technical linkages, this requires minimal harmonization of business practices and technical procedures. The majority of securities settlement systems and most clearinghouses in Europe are linked to each other, either directly or indirectly, resulting in hundreds of links (the so-called spaghetti model). However, this model has been widely criticized by market participants for failing to deliver the anticipated cost reduction and settlement efficiency and failing to allow efficient and cost-effective provision of value-added services.

Through mergers and alliances, existing securities infrastructures can be consolidated, either horizontally or vertically. At the national level, horizontal consolidation of clearing and settlement systems is largely complete. Consolidation was fueled not only by mergers between trading platforms but also by efforts to consolidate post-trade systems across financial instruments in order to reduce total settlement costs. In particular, many countries merged systems for settling equities with systems for settling government bonds that were formerly managed by central banks. In some countries, vertical consolidation occurred as the stock exchange merged with the CCP and the securities settlement system to create a securities infrastructure “silo” for equities, debt instruments, and derivatives. Deutsche Börse was the first group to integrate all these functions when it became the only shareholder of the companies providing post-trading activities. Market participants were technically forced to use these facilities when trading on its stock and derivatives markets. Borsa Italiana and Bolsas y Mercados Españoles later followed the same model. The vertical silos are generally seen as a brake on competition and an impediment to cross-border integration.

At the European level, horizontal consolidation is driven primarily by the stock exchanges. Securities settlement systems in Euronext countries merged into Euroclear,44 which also serves the Irish markets, whereas Euronext-related CCPs merged into LCH. Clearnet, which also serves MTS markets.45 Securities settlement system Deutsche Börse Clearing (Germany) merged with Cedel (Luxembourg) to create Clearstream International, whereas Eurex clearing was created as a CCP for the German and Swiss markets. In the Nordics, the integration of trading infrastructures within the OMX group has not been matched by that of clearing and settlement systems, although the Nordic Central Securities Depository (NCDS) was formed by merging the Swedish and Finnish CSDs.46 So far, cross-border consolidation has mainly taken place at the level of ownership and not by integrating infrastructures into single systems. This means that the anticipated cost savings have not yet materialized.

Securities Infrastructure Integration in Europe

There is no consensus among market players and service providers as to the best integration model for Europe. Custodian banks would like to see a single, user-owned securities infrastructure, similar to the DTCC, serving all European securities markets but providing only basic clearing and settlement without value-added services. Service providers are concerned that their past investments may lose value if incompatible cross-border solutions are chosen and that their (often highly profitable) market positions will be eroded. In particular, service providers in smaller financial markets would be satisfied with the integration of procedures without further consolidation. Even among users, views differ. Large cross-border banks have a clear interest in consolidated cross-border infrastructure. Smaller, local investment firms and banks, including local custodian banks, that risk seeing their role reduced, prefer local infrastructure.

The uneven distribution of potential costs and benefits has made clearing and settlement reform a controversial topic. National policymakers have been closely involved in the reform debate, generally defending the interests of their national players and resisting the closure of national systems. The involvement and strong views of national authorities render the outcome of any EU-level legislative initiatives uncertain. However, a clear view on the desirable integration model has not yet been articulated at this level. While the ECB and the European Commission have been encouraging integration, until recently they were reluctant to take on a leadership role or provide guidance in any particular direction. Since the summer of 2006, though, there have been some breakthroughs.

Recent Progress

The Commission has generally favored market-based solutions and has been pursuing a multipronged approach. It established an advisory and monitoring group (CESAME) to assist the private sector, did preparatory work on legislation to tackle barriers of a public-policy nature and open up markets, and set up working groups to address legal and tax issues. It is also seeking to ensure the effective implementation of competition law. For the time being, it has decided against a directive to open up markets, as it succeeded in convincing the industry to agree to a voluntary Code of Practice, which was signed on November 7, 2006. If fully implemented, the code would achieve most of the objectives the Commission would have aimed for with a directive. The main objectives of the code are price transparency, interoperability and effective rights of access on a fair and nondiscriminatory basis, and unbundling of services and accounting separation. The code initially applies only to cash equities and will be implemented in phases between end-2006 and January 1, 2008.

In July 2006, the ECB announced that the Eurosystem intended to explore the possibility of setting up a securities settlement service for securities transactions in central bank money. The facility, called TARGET2-Securities (T2S), would be integrated with TARGET2, the second generation of the Eurosystem payment system, to process both the securities and the cash legs of settlement operations on a single platform through common procedures (see ECB, 2006c). After conducting a series of public consultations, which yielded support from users but opposition from service providers, producing a number of feasibility studies, and publishing a blueprint (ECB, 2007a), the Governing Council of the ECB decided on March 8, 2007, to go ahead with the first phase of the project (the definition of user requirements). A decision on whether to proceed with the development phase is due by early-2008. As planned, T2S would introduce a single technical platform that would maintain the securities accounts of multiple CSDs, as well as a settlement engine that would allow direct settlement between these securities accounts and TARGET2 cash accounts. Putting in place T2S is expected to take 4–6 years. The Eurosystem’s provisional planning outlined in the blueprint foresees the completion of the project by the first quarter of 2013.

The implications of these initiatives are not yet fully clear. Much will depend on the full and correct implementation of the Code of Conduct by market participants. The Commission’s previous attempts to achieve progress in various aspects of financial integration by means of moral suasion, for example in the field of retail payments, have tended not to yield the desired results. The code is a novel approach, though, which may create greater pressure toward compliance. The initial signs are encouraging, as some market participants have announced their intention to voluntarily apply the code to securities other than cash equities. For its part, T2S would alter the entire securities clearing and settlement landscape in Europe. As a single cross-border system that has an implicit central bank guarantee and reduces costs as well as credit and liquidity risk by settling in central bank money, it would attract the lion’s share of clearing and settlement in the euro area market. The role of domestic systems would be reduced to depository functions and settling transactions in securities that target retail domestic investors and/or have a relatively low liquidity.

However, neither the code nor T2S will be the end of the clearing and settlement integration road. Coverage of the code would have to be extended to instruments other than cash equities for its benefits to become more universal, and a number of problems remain unaddressed. In this regard, the code calls upon European and national authorities to continue working toward the elimination of the remaining Giovannini barriers, in particular legal and tax issues, and toward supervisory convergence. Legislative initiatives by the Commission may still be needed to address these issues.

The consolidation of clearing and settlement infrastructures also implies increased cross-border systemic risks. The current decentralized prudential framework, which is based on the home country principle, may not be effective in addressing these risks, and a more centralized framework may be needed to oversee large cross-border systems (see Kazarian, 2006).

1Efforts to integrate the European Union’s equity markets through regulatory harmonization and mutual recognition started in earnest during the late 1970s and the 1980s. These efforts comprised directives that sought to harmonize conditions for the admission of securities to official stock exchanges (Directive 79/279/EEC), prospectus publishing requirements (Directive 89/298/EEC), and rules with respect to insider dealings (Directive 89/592/EEC), as well as a directive to introduce mutual recognition for “undertakings for the collective investment in transferable securities” (UCITS, a harmonized kind of open-ended collective investment fund) (Directive 85/611/EEC). However, by far the most important initiative in the pre-FSAP era was the Investment Services Directive (Directive 93/22/EEC), which came into force in 1996 and sought to introduce a single passport for investment firms as well as harmonize regulations for securities markets. For an overview, see Licht (1997).
2The Portuguese stock market joined in 2002, and the Euronext group later bought the London International Financial Futures and Options Exchange (LIFFE). In late 2006, Euronext NV and NYSE Group, Inc., the operator of the New York Stock Exchange, finalized a merger agreement. In early 2007, they established a holding company, NYSE Euronext. Inc. that is intended to become the sole owner of both stock market groups. The U.S. and European exchanges will continue to function alongside each other and be regulated by national authorities, following the decentralized model Euronext has developed at the EU level.
3HEX had previously acquired the Latvian and Estonian exchanges, acquired the Lithuanian stock exchange, merged with the Copenhagen stock exchange (CSE), and bought a stake in the Oslo exchange.
4Deutsche Borse AG operates a range of trading institutions out of Frankfurt, including trading in cash securities on the Frankfurt Stock Exchange; trading in futures and options through the Eurex market, which it co-owns with the Swiss exchange; settlement and clearing services through its Clearstream subsidiary; and market information through the DAX index.
5Data based on the dispersion of (HP) filtered total stock market returns reported in the ECB’s “Indicators of financial integration in the euro area”; see the Statistics tab on the ECB’s website (www.ecb.int). A second indicator provided there is based on the proportion of variance in local equity returns explained by euro area versus U.S. shocks.
6Fratzscher (2001) and Adjaouté and Danthine (2001) find increased correlation among EMU stock returns. However, Adjaouté and Danthine (2004a) find instability in correlation measures from their previous study.
7To control for inherent differences in the underlying asset(s), Levy-Yeyati, Schmukler, and Van Horen (2006) examine the “cross-market premium” or price differential of the same, cross-listed stock on multiple exchanges as an arbitrage-based measure of financial integration.
8From the perspective of cash flow analysis, equity market integration should affect the choice of discount rate but not the underlying behavior of the cash flows (which instead depend on real factors such as economic integration). Drawing on Chen and Knez (1995), Sontchik (2003) similarly argues that under financial integration the “same stochastic discount factor is used to value uncertain but identical cash flows … assets with identical risk should command the same expected return regardless of location.”
9Dumas, Harvey, and Ruiz (2003) theoretically link stock market correlations to countries’ output correlations.
10According to the Krugman-Venables hypothesis, greater regional specialization at the sectoral level may imply declining similarity in the composition of geographically oriented national stock market indices under EMU; Kalemli-Ozcan, Sorensen, and Yosha (2001, 2003) find that industrial specialization has intensified within the European Union.
11Brooks and Del Negro (2002a, 2002b, 2002c), for example, explore the extent to which the ICT bubble has been responsible for the global rise in stock market comovements and the increasing importance of industry over country factors. Adjaouté and Danthine (2001) interpret increasing correlations among EU-15 stock markets as increasing “homogenization” of shocks rather than integration. See also Adam and others (2002).
12Data are from Datastream and are expressed in U.S. dollars. Data on total returns from Bloomberg (expressed in local currency terms) are used as a cross-check after currency conversion.
13National stock market indices used from Datastream and Bloomberg were as follows: ATX (Austria), BEL 20 (Belgium), KFX (Denmark), HEX (Finland), CAC 40 (France), DAX (Germany), ASE (Greece), ISEQ (Ireland), MIB (Italy), BVLX (Portugal), AEX (Netherlands), IBEX (Spain), OMX (Sweden), and FTSE 100 (United Kingdom). As an alternative to narrower market indices, broader-based measures—such as the United Kingdom’s FTSE 350 or France’s SBF 250—were also used to cross-check the returns data. Ireland was later dropped as an origin country in the subsequent analysis due to lack of data observations.
14The issue of common shocks affects not only the interpretation of the extent to which simple correlations can be seen to measure integration but also has a bearing on diversification issues examined here. Moerman (2004), for example, finds that sector-based diversification outperformed country-based strategies during 1995–2002 (a period covering the ICT boom and bust), but Ehling and Ramos (2005) find that geographical diversification still outperforms in the presence of portfolio constraints (namely, no short-selling). In what follows, the analysis explores the possible benefits of country diversification within the European Union.
15In addition to the normality test statistics reported in the table, Jarque-Bera and Kolmogorov-Smirnoff tests find broadly similar results, albeit with fewer departures from normalcy in the latter instance.
16Note that the risk-return frontier and actual portfolio holdings are best conceptualized as being simultaneously determined at a given point in time. Some care must be taken when considering changes in portfolio holdings to “move toward” a given frontier, as portfolio changes would affect stock prices and returns, and hence the location of the frontier itself. Basak, Jagannathan, and Sun (2002) propose a simpler mean-variance efficiency test based on “horizontal distance” to the frontier in Figure 5.3. Specifically, for a sample portfolio with expected return μ and variance σ2, the measure is defined as the difference λ = Var (μ) – σ2, where Var (μ) is the risk of the efficient portfolio with the same expected return μ.
17Numerous studies have examined the puzzle of home bias, including French and Poterba (1991); Tesar and Werner (1995); and Baxter and Jermann (1997). See Lewis (1999) for a survey.
18Examining gross, cross-border equity flows in a gravity-equation framework, Portes and Rey (2005), for example, find a key role for informational asymmetries (proxied by distance) in explaining the pattern of asset trade.
19Kho, Stulz, and Warnock (2006) argue that despite financial globalization, home bias remains stubbornly high, in part because of the role of company insiders who hold significant shares of company stock due to corporate governance issues.
20The relevant distributional condition is actually ellipticality (normality is a special case). Meyer (1987) demonstrates that comparative statistics using the mean-variance approach is valid for any utility function when returns are jointly elliptical.
21Relevance of higher moments for the second-order approximation can be illustrated as follows. If consumption is a function of financial returns with mean return μ and variance σ2, expected utility can be written using the following Taylor expansion:E[u(c)]=u(μ)+u(μ)σ22!+u(μ)E[(cμ)3]3!+...Three cases for the sufficiency of risk and return are pertinent: (1) If utility is quadratic, u′′′ and higher derivatives are zero; or (2) if stochastic returns are elliptical, then higher moments are monotonic functions of first and second moments—that is, indirect utility can be written as v(μ,σ) = E[u(c)]; ∂v/∂μ >0, ∂v/∂σ < 0; or (3) if higher moments yield small welfare consequences, then a second-order approximation is appropriate.
22Michaud (1989) argues that the mean-variance approach overweights assets with large estimated returns, low variance, and negative correlations that may reflect estimation errors—more likely in the case of such assets.
23The literature has proposed various alternatives to improve estimation of the covariance matrix of stock returns. See Chan, Karceski, and Lakonishok (1999); Bengtsson and Holst (2002); Jagannathan and Ma (2003); Ledoit and Wolf (2003, 2004a, 2004b); and Disatnik and Benninga (2005) for alternative estimators to address estimation risk. Volatility clustering (time-varying variances such as ARCH or GARCH processes) is a related issue, raising additional measurement issues. In the analysis, expanding and rolling windows for calculating risk and return are used to account for this.
24To address this, Jagannathan and Ma (2003) and Black and Litterman (1992) constrain portfolio strategies by excluding short-selling to reduce estimation risk and to achieve more balanced portfolio positions. Green and Hollifield (1992), however, argue that short-selling may be justified on economic grounds (if factor risk dominates estimation risk). Specifically, if the covariance of returns is heavily influenced by a common factor, it may be optimal to take short (negative) positions in some assets to fund long positions in others (with lower volatility) to reduce systemic risk.
25The numerical value of the diversification cone is also difficult to interpret. Therefore, in what follows, the “distance” is expressed in relative terms—normalized compared to the efficiency of the EU uniform portfolio (an interior point) and to also facilitate crosscountry comparisons.
26This is not to say that other dynamics—namely, financial development and technological advance—are not also at work. Financial development and integration may also be complementary forces, as discussed in Chapter 3.
27See IMF (2002) for a full description of the scope, concepts, methodology, and modalities of the survey.
28The 1997 survey covered 29 countries, including all 15 EU member states, but Greece, Germany, and Luxembourg were recorded only as destination, not source, countries.
29As discussed in IMF (2002), the CPIS is confronted with numerous methodological and measurement issues, including determining country of attribution from surveys of investment managers, trusts, and custodians where the beneficial owners are not always easily identified. This tends to bias upward the holdings data of countries with important financial centers.
30Alternative estimates of home bias using other data-based definitions are discussed in Appendix 5.1. In the following section, home bias is discussed from a perspective based on economic theory.
31See Appendix 5.1 for a comparison of portfolio efficiency in 1997 and in 2003 using the uniform portfolio benchmark in each respective year.
32In the figure, the global minimum variance (GMV) portfolio denotes the frontier point that attains the lowest portfolio variance possible for EU-15 equities based on the behavior of historical returns over the sample.
33Table 5.6 shows the percent change (i.e., log difference) in the diversification cone or area below the respective efficient frontiers in 1997 and 2003. Note that the efficiency of the uniform EU portfolio with respect to the frontier also changes over time; this effect is taken into account in the table. Appendix Table A5.2 shows the individual estimates of relative diversification efficiency in years 1997 and 2003, for side-by-side comparison, but subject to this caveat regarding the benchmark’s efficiency rating.
34This is analogous to concern about trade diversion, when trade patterns are altered among trading partners forming regional or bilateral trade alliances such as customs unions or free trade areas.
35For clearing and settlement, the Giovannini Group identified 15 barriers in its 2001 report. In a follow-up report, precise actions were identified for removing barriers related to the harmonization of operating hours and settlement deadlines, the guarantee of intraday settlement finality in all links between settlement systems, and the removal of practical impediments to remote access to national clearing and settlement systems. See also Schmiedel and Schönenberger (2005).
36See, for example, the ECB “blue book” (2001).
37CESAME is a group of advisers brought together by the European Commission to assist it and the private sector in removing those Giovannini barriers that are under the control of the private sector.
38In this regard, seven leading investment banks (Citigroup, Credit Suisse, Deutsche Bank, Goldman Sachs, Merrill Lynch, Morgan Stanley, and UBS) plan to launch a pan-European trading platform designed to reduce the costs of buying and selling shares when MiFID becomes effective in the fall of 2007, taking advantage of the directive’s removal of the rule that shares can be traded only on a stock exchange (and its designated trading infrastructure).
39According to the IMF’s Balance of Payments Manual (5th edition, 1993), a direct investment is defined when “a direct investor, who is resident in another economy, owns 10 percent or more of ordinary shares or voting power.” See also IMF (2002).
40This assumption introduces a bias or distortion to the estimates, but on the other side, namely, by understating Luxembourg’s true foreign equity holdings.
41Green and Hollifield (1992), for example, argue that short-selling can reflect factor risk. If stock returns—often highly positively correlated—are driven by a dominant factor (e.g., tech boom), then optimal portfolio diversification will try to reduce factor risk by taking large negative positions in (say) one index to finance (larger) positions in another.
43Prepared by Elias Kazarian and draws on work published in Kazarian (2006).
44The entities that merged to form Euroclear were Euroclear Bank (an ICSD), Sicovam (France), Necigef (Netherlands), CrestCo (United Kingdom), and CIK (Belgium).
45LCH. Clearnet resulted from the merger between the London Clearing House (United Kingdom) and Clearnet (France). MTS is the leading market in Europe for the trading of fixed-income securities.
46Attempts to create a Nordic unified clearing and settlement system have thus far failed.

    Other Resources Citing This Publication