EDHEC-Risk Days 2012 Singapore, 9-10 May 2012

Transcription

1

2 EDHEC-Risk Days 2012 Singapore, 9-10 May 2012 Assessing the Quality of the Major EquityIndices in Asia Felix Goltz, PhD Head of Applied Research, EDHEC-Risk Institute

3 Quality of Existing Indices Setting the context for this study A flurry of recent academic literature has clearly established that vast majority of investor s return is attributed to investor s choice of asset allocation, and not due to security selection or market timing (Brinson et al. 1986, Ibbotson et al. 2000, etc) In order to make an asset allocation decision investors use indices as a proxy for portfolios that reflect their choice of allocation In equity markets, the choices of these indices are usually the standard indices available in the market like S&P 500, Wilshire 5000, FTSE 100 etc. In this research we look into the quality issues of choosing such standard indices which almost always happen to be capitalization weight indices 3

4 Quality of Existing Indices in Asia Setting the context for this study In the current research, we look into the following three key quality criteria of an index: Risk/Return Efficiency and Concentration in the Indices Stability to Style and Sector exposures Economic Representativity This research is an extension of our previous analysis for US/European Indices based on Amencet al (2006) with the focus being on Asian Indices The indices are chosen from major Asian equity markets and are chosen based on their popularity and usage in terms of indexing/ benchmarking of funds in the region. Indices chosen for the analysis Index Hang Seng Index Nikkei 225 Index Topix 100 Index FTSE STI Index Kospi 200 Index FTSE TWSE 50 Index CSI 300 Index FTSE China 25 Index Nifty 50 Index FTSE ASEAN Index Country Hong Kong Japan Japan Singapore Korea Taiwan China China 1 India ASEAN Region 2 1-The index consists of the largest 25 Chinese stocks listed and trading in Hong Kong Stock Exchange. Hence the index base trading currency is Hong Kong Dollar. 2-FTSE ASEAN Index consists of stocks from Singapore, Malaysia, Indonesia, Thailand and Philippines which are a part of the ASEAN block of countries. 4

5 Quality of Existing Indices in Asia Risk/Return Efficiency Concentration effects Stability to style and sector exposures Representativeness 5

6 Efficiency test of indices: Previous evidence Lack of Mean-Variance Efficiency The standard indices weight stocks proportional to their market capitalization This belief is based on the Capital Asset Pricing Model(CAPM). The CAPM assumes that each investor holds the same efficient tangency portfolio, and therefore concludes that the aggregate portfolio held by investors (which by definition is cap weighted) is also efficient Empirical findings(haugen and Baker 1991, Grinold 1992, Schwartz 2000) offer evidence that market-cap weighted indices fail to provide an efficient risk-return profile. Haugen and Baker (1991): Cap-weighted stock portfolios are inefficient investments.. Even the most comprehensive cap-weighted portfolios occupy positions inside the efficient set. Cochrane (2001): Market indices [ ] are if anything inside that [mean-variance] frontier 6

7 Efficiency test of indices: Previous evidence Index Inefficiency - Empirical Findings Cap-weighted index lies deep inside the ex-post efficient frontier. 7 Based on data for the period The efficient frontier assumes a perfect forecast of the future covariance matrix and of the future mean return. Figure taken from Schwartz (2000), Figure 3, page 19.

8 Efficiency test of indices: Previous evidence Index Inefficiency - Theoretical Reasons It is not surprising that market cap weighted indices are not efficient portfolios. When relaxing the highly unrealistic assumptions of the CAPM (such as identical preferences and time horizons, frictionless markets), financial theory does not predict that the market portfolio is efficient(sharpe(1991), Markowitz(2005)). Besides, even if the world worked according to the CAPM theory, standard equity indices do not represent the true market portfolio (which reflects all wealth in the economy as opposed to only the major stocks). 8

9 Efficiency test of indices: Previous evidence Index Inefficiency Inefficient risk/reward ratio It is possible to construct a portfolio with higher risk/reward ratios Haugen and Baker (1991) test the Wilshire 5000 index, which is the most comprehensive cap-weighted index in U.S and suggest that it was possible to construct equity portfolios with higher risk/reward efficiency than their cap-weighted counterparts Sinclair (1998) also supports the previous theoretical arguments that investment opportunities existed to build equity portfolios that exhibit better risk/reward ratios than cap-weighted indices. Grinold(1992) examine five equity benchmarks in the US, the UK, Australia, Japan and Germany and show that out of these five countries, the first four indices are not efficient which implies that other portfolios could be constructed to outperform the benchmarks. Cap-weighted perform worse than non-cap weighted portfolios! Amencet al. (2011), comparing alternative weighted indices like fundamental index, equal weighted index, efficient index, and minimum volatility index, find that all these indices show on an average returns superior to those of cap-weighted equity indices. Platen and Rendek(2010) observe that the equal weighted portfolios constructed from country indices in each country had higher Sharpe ratios than the corresponding Cap weighted indices in all 53 countries tested. Tamura & Shimuzu(2005) and Chou et al. (2006) also show that the characteristics based portfolio produce perform better over the cap weighted indices 9

10 Efficiency test of indices: Previous evidence Cap Weighted versus Equally-Weighted Portfolios Extracted from Platen and Rendek(2010). Differences in Sharpe ratios between 53 EW country indices and their CW counterpart. Sample period is January 1973 until March

11 Efficiency test of the Asian market indices Efficiency Analysis Data and Methodology In the first part of this research on the quality of indices we test the mean variance efficiency for a set of market indices We test the efficiency of these market indices by building the efficient frontier in sample, and identify how far inside from the efficient frontier is the market index portfolio. We also locate the position of the equally weighted portfolio in this plane to test it s relative efficiency with respect to the market index We obtain comparability of our optimised portfolios with cap-weighted indices The data of indices are adjusted for all the historical change to the constituents, and hence completely free of survivorship bias In our approach to building the frontier and the EW portfolios we maintain the same representativity of stock constituents as that of the market index at all time through the sample 11

12 Efficiency test of the Asian market indices Results We report the measure of inefficiency by using the Graham Harvey Measure score, which is the difference in returns between a reference portfolio (here the Equal weighted and the efficiency portfolios) and the market index portfolio, after the reference portfolio s volatility has been scaled to match the market index volatility. Market Index Improvements in risk-adjusted annual performance through Equal-Weighted, Maximum Sharpe Ratio and Minimum Variance Indices compared to Standard Index Time Period Averageannual market return GH2 Measure of GH2 Measure of Max GH2 Measure of EW Portfolio over Sharpe portfolio MinVar portfolio mkt index over market index over mkt index Hang Seng Jan Dec % 2.54% 28.61% 6.75% NIKKEI 225 Jan Dec % 4.81% 41.37% 6.43% TOPIX 100 FTSE STI Feb Dec 2010 Sep Dec % 5.79% 37.52% 5.48% 15.4% 6.52% 30.20% 7.24% KOSPI 200 Jun Dec % 5.72% 53.58% 6.52% TWSE 50 Jan Dec % 2.22% 33.80% 6.99% CSI 300 Jan Dec % 16.44% 57.92% 20.52% FTSE China 25 Jan Dec % 4.26% 32.97% -5.68% NIFTY 50 Jan Dec % 6.29% 42.98% 10.22% 12 FTSE ASEAN Jan Dec % 2.71% 38.45% 3.25%

13 Efficiency test of the Asian market indices Results Clearly all indices are highly mean variance inefficient as evident from their GHM scores The inefficiency of cap weighted is across the board and pervasive The extent of inefficiency is slightly lower in the case of developed countries in Asia (by the measure of relative performance against maximum Sharpe ratio portfolio) Further, in all instances the Equal weighted portfolio outperforms the cap/price weighted indices by at least 200 basis points We pick out the performance of a set of developed countries in Asia and see its efficient frontier, as some these also happen to be most efficient indices based on their relative GHM score against the maximum Sharpe ratio portfolio Also, given the long time series history we have on Nikkei, we see the annual efficiency plots of the index during different market regimes 13 like 1998to2001,2007to2009etc.

14 Efficiency test of the Asian market indices Inefficiency in developed Asian markets Example of Hang Seng Equally weighted portfolio Hang Seng Index 14

15 Efficiency test of the Asian market indices Inefficiency in developed Asian markets Example of Nikkei Equally weighted portfolio Nikkei 225 Index 15

16 Quality of Existing Indices in Asia Market Efficiency Concentration effects Stability to style and sector exposures Representativeness 16

17 Concentration effects in the Asian indices Motivation One of the reasons why Cap-weighting is particularly inefficient is because it leads to high concentration: the effective number of stocks in the index is low. The effective number of stocks is the reciprocal of the Herfindhal index, a measure of portfolio concentration Index Nominal number Effective number S&P NASDAQ FTSE 100 (UK) FTSE Eurobloc FTSE Japan Average effective number based on quarterly assessment for the time period 01/1959 to 12/2008 for the S&P, 01/1975 to 12/2008 for the NASDAQ, and 12/2002 to 12/2008 for the other indices. 17

18 Time Concentration effects in the Asian indices Literature review Malevergneet al. (2009) argue that cap-weighted indices are in general heavily concentrated in a few large firms Malevergneand Sornette(2007) followed by Malvergneet al. (2009) also argue that there is a new source of risk that should be priced into the assets in addition to market portfolio when the portfolios are highly concentrated in few stocks. Goltzand Sahoo(2011) presents simplified examples of the negative effects of concentration on performance, and how it produces significant drag in market portfolio returns due to relative underperformance of a single large stock in the index. Impact of Single Stock Behaviour on Cap-weighted Indices 2010Q2-2010Q3 Name FTSE100 Cap-weighted Index Return Volatilit y Name -6.70% 20.20% British Petroleum 10/29/2008 DAX -0.30% N.A. Volkswagen 2000Q3-2001Q3 STOXX 50 Return Single Stock % % Volatilit y Weight at the peak Index with relevant stock down weighted to a 1/N weight Return Volatility 47.10% 8.00% -0.30% 20.80% N.A % 15.90% N.A % 23.60% Nokia % 73.00% 10.50% % %

19 Concentration effects in the Asian indices Results Table below is the summary of the results from the concentration analysis. Column 4 represents the effective number of stocks which is the reciprocal of Herfindahlindex.. In column 5 we calculate the effective measure, which is calculated as the ratio of Effective number of stocks/ Number of constituents in the index. Index Time Period Concentration in standard market indices Average nominal number of constituents Average effective number of stocks (Effective number of stocks)/(nominal number of stocks) Weight concentration in top 20% of constituents Hang Seng Jan 2002 to Dec % 63.3% Nikkei 225 Feb 2001 to Dec % 61.2% Topix 100 Jan 2001 to Dec % 49.7% FTSE STI Feb 2008 to Dec % 50.3% Kospi 200 Feb 2002 to Dec % 80.0% TWSE 50 Jul 2003 to Dec % 52.3% CSI 300 Sep 2005 to Dec % 59.5% FTSE China 25 Mar 2004 to Dec % 40.2% Nifty 50 Feb 2002 to Dec % 56.7% 19 FTSE ASEAN Jan 2001 to Dec % 63.3%

20 Concentration effects in the Asian indices Results/ Takeaways As we see from the above table there is a strong concentration effect in the Asian Indices The effective number of stocks are less than half the total number of stocks in almost all cases Couple of interesting facts to note are: The concentration in FTSE China 25 index is the least, as the weights are capped by construction Kospi 200 index is one of the most concentrated indices due to presence of stocks like Samsung that occupy close to 18% of the index 20

21 Quality of Existing Indices in Asia Market efficiency Concentration effects Stability to style and sector exposures Representativeness 21

22 Factor stability test of the Asian indices Factor exposure Literature review in Asian Context Investors typically view the risk exposures of their investment universe by breaking it down into simpler blocks based on style, sector or size based factors In this setting, a market index chosen for passive investment may or may not be neutral to these factor exposures over time In fact, there can be significant variation in these factor exposures for the standard equity indices over time Bahriand Leger (2001) have studied the stability of risk factors in the UK stock market over time by running multi-period PCA on 550 stocks from January 1972 to December They find that it is not reasonable to assume constant risk factor exposure for the stock market. Amencet al (2006) observe that the variation in factor weights on style and sector exposures are significant, on an analysis done on select set of equity indices in US, Developed Europe and Japan Goltzand Sahoo(2011) also show that the evolution of sector weights of S&P500 from January 1959 to December 2008 has a shift from a manufacturing industry base towards information technology and the financial services industry. In this final section of our analysis, we identify the extent of variation in factor weights with respect to styles and sectors, to understand the Asian indices relative stability to these risk exposures 22

23 Stability of Sector exposures Methodology We identify the variation in the sector exposure using a holdings based analysis We added up the weights of stocks within one sector (defined by GICS) and plotted the evolution of individual sector weights within the indices over time. This helps us visualisethe extent of variation in weights over time. The plots of the individual sector weights over time are presented in the appendix Further, to summarize our results we also calculate the sector drift score, and note the dispersion in weights of the largest sector in the index The sector drift score is the average variation of all sector weights. Lower the value of this score the more stable are the sector weights within the index The formula for calculation of the score is as below 23

24 Stability of sector exposures Results Summary of the sector drift score based on sector exposure. Geographical zone Hong Kong Index Period Max Weight Change Sector drift Sector Max. Weight Min. Weight score Hang Seng Index Jan 2001 to Dec 2010 Telecommunication Services 27.68% 8.56% 8.89% Japan NIKKEI 225 Index TOPIX 100 Index Singapore FTSE Strait Times Index South Korea KOSPI 200 Index Taiwan FTSE TWSE Taiwan 50 Index China CSI 300 Index India ASEAN FTSE China 25 Index NIFTY Index FTSE ASEAN Index Jan 2001 to Dec 2010 Jan 2001 to Dec 2010 Feb 2008 to Dec 2010 Feb 2002 to Dec 2010 Jul 2003 to Dec 2010 May 2005 to Dec 2010 May 2005 to Dec 2010 Jan 2002 to Dec 2010 Jan 2001 to Dec 2010 Information Technologies 33.15% 17.40% 4.83% Financials 29.57% 12.38% 6.81% Financials 53.61% 43.44% 4.39% Industrials 28.57% 7.91% 8.44% Information Technologies 62.38% 46.65% 6.51% Financials 44.12% 13.43% 11.44% Financials 51.07% 17.19% 11.11% Consumer Staples 27.54% 3.08% 11.51% 24 Financials 45.83% 36.18% 4.94%

25 Stability of sector exposures Example of the evolution of sector weighting For the index with the lowest drift score * 100% Evolution of sector weights of NIKKEI 225 Index from 2001 to 2010 (drift score: 4.83%) 90% 80% 70% 60% 50% 40% 30% 20% 10% Utilities Telecommunication Services Information Technology Financials Health Care Consumer Staples Consumer Discretionary Industrials Materials Energy 0% Though FTSE STI Index has the lowest drift score of 4.39%, the sample period is only about 3 years which is not comparable with other indices which have sample period at least last for 6 years.

26 Stability of sector exposures Example of the evolution of sector weighting For the index with the highest drift score 100% Evolution of sector weights of Nifty 50 Index from 2002 to 2010 (drift score: 11.51%) 90% 80% 70% Utilities Telecommunication Services 60% Information Technology Financials 50% Health Care 40% Consumer Staples Consumer Discretionary 30% Industrials 20% Materials Energy 10% 0% 26

27 Stability of sector exposures Results The sector drift scores for indices in developing countries, such as China and India, are much higher than the indices in Europe, Japan and US from Amencet al. (2006) study*. Indices in Asia developed markets, such as Hong Kong, Japan, Singapore, South Korea and Taiwan, demonstrate better sector factor exposure stability than indices in developing markets Financial crisis from end of 2007 to 2009 also contributes in the sector weight variations. However, even during the relative calm periods, the variation in sector weights does occur for all indices. Note that the sector weights will change with the price even if there is no constituents change as the index weights stocks by their market capitalisations or prices. Investors are exposed to implicit sector exposure chosen by the market index and such exposures are varying over time. 27 The average sector drift scores for FTSE All Share Index 700, DJ Euro Stoxx300, DJ Stoxx600, Topix 1666 and S&P 500 are about 6.5 to 7.5%, except Germany Prime All Share Index 380, which is 10.4%.

28 Stability of style exposures Methodology We analyze the variation in the style weights using a returns based analysis by identifying the evolution of the individual style weights over time. We regress the country index returns on the corresponding MSCI country value/growth index returns where we force the factor weights on each factor to be between zero and 1, and such that the sum of the weights add up to 1 This helps us visually see the extent of variation in weights over time. The plots of the variation in weights of the factors are presented in the appendix In this analysis, we exclude the FTSE ASEAN index as there is no corresponding MSCI style index. In addition, for CSI 300 index, we use MSCI China A style indices as they both represent the A-share market. Whereas, for FTSE China 25, we use MSCI China style indices, which composes share classes traded by foreign investors. Further, to summarize our results we also calculate the style drift score, similar to our approach for evaluating the sector drift score, and as defined by Idozerk and Bertsch(2004) As defined before, the style drift score is the average variation of the style weights. Lower the value of this score the more stable are the style weights within the index 28

29 Stability of indices to style exposures Results Summary of the style drift score based on style exposure. Geographic zone Hong Kong Hang Seng Index Japan Singapore South Korea Taiwan Index Period Max. Weight (value) Jan 2002 to Dec 2010 NIKKEI 225 Index Jan 2002 to Dec 2010 TOPIX 100 Index Jan 2002 to Dec 2010 FTSE Strait Times Jan 2002 to Dec Index 2010 KOSPI 200 Index Jan 2002 to Dec 2010 FTSE TWSE Taiwan 50 Index Jan 2002 to Dec 2010 China CSI 300 Index Jan 2003 to Dec 2010 FTSE China 25 Mar 2002 to Dec Index 2010 India NIFTY Index Jan 2002 to Dec 2010 Min. Weight (value) Style drift score Index return volatility (p.a.) 53.04% 16.55% 11.77% 25.68% 57.75% 0% 20.59% 25.32% 70.98% 35.64% 12.79% 24.67% 56.15% 33.39% 6.39% 20.38% 70.53% 46.87% 8.48% 25.91% 73.83% 56.01% 5.80% 22.81% 72.68% 44.28% 8.96% 29.99% 88.96% 45.37% 19.53% 31.85% 67.18% 46.31% 6.64% 26.82% 29

30 Stability of indices to style exposures Results For the index with the lowest drift score Evolution of style exposure for FTSE TWSE Taiwan 50 Index from 2002 to 2010 (drift score: 5.8%) 100% 90% 80% 70% 60% 50% 40% 30% Growth Value 20% 10% 0% 30

31 Stability of indices to style exposures Results For the index with the highest drift score 100% 90% 80% 70% 60% Evolution of style exposure for Nikkei 225 Index from 2002 to 2010 (drift score: 20.59%) 50% 40% 30% Growth Value 20% 10% 0% 31

32 Stability of indices to style exposures Results All indices show considerably variability of style exposure.the range of style drift scores varies from 5.8% to 20.59%, while sector drift scores vary from 4.39% to 11.51%. The Japanese indices have the highest variation in the evolution of the style exposure, for both NIKKEI 225 Index and TOPIX 100 Index (20.59% and 12.79%, respectively). Indices in developing countries, for instance NIFTY Index, on the other hand show a lower style drift score than most of the indices from the developed countries Asian market indices have more instable style exposure than European market indices which are from 2.6% to 5.5%, shown in Amenc et al. (2006) There is no causative relation between the changes in style exposure and the annual volatility. Having an index with lower volatility does not mean that its style exposure will be less volatile 32

33 Quality of Existing Indices in Asia Market efficiency Concentration effects Stability to style and sector exposures Representativeness 33

34 Representativeness of Asian Indices Introduction Investors look to gain economic exposure to a specific country or region by investing in the respective country index Do current country indices provide such representativeness? Representing a market place vs. Representing an economy Standard country or regional indices contain companies listed in the area without necessarily having a strong economic link with the region E.g. L Occitane(HKSE 973) has been assigned Hong Kong Nationality in 2010 by FTSE as the company became listed on the Hong Kong Stock Exchange. However, L Occitanewas founded in France and incorporated in Luxembourg with 66% of revenues, 78% of PPE, and 76% of intangible assets outside of Asia-ex-Japan. 34

35 Representativeness of Asian Indices Introduction Why seeking access to the economic exposure? Acting upon economic predictions Capturing the growth of a region Managing specific hedging demands What is economic exposure? Exposure to Asian internal growth: which company is highly dependent on internal demand in Asia? Exposure to Asian success: which companies benefit most from Asian competitiveness? They can be defined either by demand (revenues) or by supply (production) Our study aims to investigate the representativeness of seven Asian indices (FTSE China 25, CSI, Hang Seng, Kospi, Nifty, TWSE and STI) based on the revenues and production figures of their constituents over the last 5 years 35

36 Representativeness of Asian Indices Geographic breakdown of Revenues Cap-weightingpredominantly leads to an increase in focus on constituents with few revenues in Asia On the other hand, equal-weighted measure tilts towards constituents which are more Asia-focused Overall, both EW and CW indices end up with an exposure of the average stock which is far from being a pure exposure to Asia. 100% AVERAGE: % of Revenues from Asia-10 --CW vs EW 90% 80% 70% 60% EW CW 50% 40%

37 Representativeness of Asian Indices Geographic breakdown of Revenues Asian country indices show pronounced differences in their Asian economic exposure. 100% % Share of Revenues from Asia-10 (CW) 90% 80% 70% 60% FTSE China 25 CSI Hang Seng Kospi Nifty TWSE STI 50% 40%

38 Representativeness of Asian Indices Geographic breakdown of Revenues The average exposure hides pronounced differences across stocks contained in the indices. However, standard index construction fails to consider such differences in exposure across stocks. AVERAGE: Distribution of constituents based on % revenues from Asia-10 AVERAGE: Distribution of constituents based on % revenues from US-EU 100% 100% 100% 100% 80% 80% 80% 80% 60% 60% 60% 60% 40% 40% 40% 40% 20% 20% 20% 20% 0% 0%-20% 20%-40% 40%-60% 60%-80% 80%-100% % of Index Weight (Market Cap) -Left Axis 0% 0% 0%-20% 20%-40% 40%-60% 60%-80% 80%-100% % of Index Weight (Market Cap) -Left Axis 0% 38 No of constituents as % of universe- Right Axis No of constituents as % of universe- Right Axis

39 Representativeness of Asian Indices Geographic breakdown of production factors When looking at production in Asia, we confirm the main insights from the Revenue-based results:. Cap-weightingpredominantly leads to an increase in focus on constituents with few production in Asia Pronounced difference sin average exposure exist across indices. AVERAGE: % of Production in Asia-10 --CW vs EW % Share of Production in Asia-10 (CW) 100% 100% 90% 80% 80% 60% 40% EW CW 70% 60% 50% 40% FTSE China 25 CSI Hang Seng Kospi Nifty 30% TWSE 20% 20% STI 10% 0% %

40 Representativeness of Asian Indices Geographic breakdown of production factors The average exposure hides pronounced differences across stocks contained in the indices. However, standard index construction fails to consider such differences in exposure across stocks. AVERAGE: Distribution of constituents based on % production in Asia-10 AVERAGE: Distribution of constituents based on % production in US-EU 60% 60% 60% 60% 40% 40% 40% 40% 20% 20% 20% 20% 0% 0%-20% 20%-40% 40%-60% 60%-80% 80%-100% 0% 0% 0%-20% 20%-40% 40%-60% 60%-80% 80%-100% 0% -20% -20% -20% -20% % of Index Weight (Market Cap) -Left Axis No of constituents as % of universe- Right Axis % of Index Weight (Market Cap) -Left Axis No of constituents as % of universe- Right Axis 40

41 Conclusion Way Forward.. Shortcomings of standard Asian indices: Standard market capitalization weighted indices are highly inefficient in terms of risk/return. These indices show severe variation in their sector and style exposure over time, thus imposing implicit risk choices upon investors. Asian indices show a lack of economic exposure, with significant differences across indices and across constituents. Outlook: Cap-weighted indices vs strategy indices: While cap-weighted indices will remain of crucial importance as peer group references and market representatives. they fall short of fulfilling investor requirements when it comes to managing risk and return If investors want to obtain risk/return efficiency, want to control their risk exposures or their economic exposure, Innovative index solutions which aim at addressing particular investor problems are necessary. 41

42 Appendix 42

43 Efficiency Plots Red dot: EW portfolio; Black dot: market index 43

44 Efficiency Plots Red dot: EW portfolio; Black dot: market index 44

45 Efficiency Plots Red dot: EW portfolio; Black dot: market index 45

46 Efficiency Plots Red dot: EW portfolio; Black dot: market index 46

47 Efficiency Plots Red dot: EW portfolio; Black dot: market index 47

48 Efficiency Plots Red dot: EW portfolio; Black dot: market index 48

49 Efficiency Plots Red dot: EW portfolio; Black dot: market index 49

50 Efficiency Plots Red dot: EW portfolio; Black dot: market index 50

51 Efficiency Plots Red dot: EW portfolio; Black dot: market index 51

52 Efficiency Plots Red dot: EW portfolio; Black dot: market index 52

53 Sector Plots Evolution of sector exposure for Hang SengIndex from 2001 to % 90% 80% 70% 60% 50% 40% 30% 20% 10% Utilities Telecommunication Services Information Technology Financials Consumer Staples Consumer Discretionary Industrials Materials Energy 0% 53

54 Sector Plots Evolution of sector exposure for Nikkei 225 Index from 2001 to % 90% 80% 70% 60% 50% 40% 30% 20% 10% Utilities Telecommunication Services Information Technology Financials Health Care Consumer Staples Consumer Discretionary Industrials Materials Energy 0% 54

55 Sector Plots Evolution of sector exposure for Topix 100 Index from 2001 to % 90% 80% 70% 60% 50% 40% 30% 20% 10% Utilities Telecommunication Services Information Technology Financials Health Care Consumer Staples Consumer Discretionary Industrials Materials Energy 0% 55

56 Sector Plots Evolution of sector exposure for FTSE STI Index from 2008 to % 90% 80% 70% 60% 50% 40% 30% Telecommunication Services Financials Consumer Staples Consumer Discretionary Industrials 20% 10% 0% 1/2/2008 1/8/2008 1/2/2009 1/8/2009 1/2/2010 1/8/

57 Sector Plots Evolution of sector exposure for Kospi 200 Index from 2002 to % 90% 80% 70% 60% 50% 40% 30% 20% 10% Utilities Telecommunication Services Information Technology Financials Health Care Consumer Staples Consumer Discretionary Industrials Materials Energy 0% 57

58 Sector Plots Evolution of sector exposure for FTSE TWSE 50 Index from 2003 to % 90% 80% 70% 60% 50% 40% 30% 20% Telecommunication Services Information Technology Financials Consumer Staples Consumer Discretionary Industrials Materials Energy 10% 0% 1/7/2003 1/7/2004 1/7/2005 1/7/2006 1/7/2007 1/7/2008 1/7/2009 1/7/

59 Sector Plots Evolution of sector exposure for CSI 300 Index from 2005 to % 90% 80% 70% 60% 50% 40% 30% 20% 10% Utilities Telecommunication Services Information Technology Financials Health Care Consumer Staples Consumer Discretionary Industrials Materials Energy 0% 1/5/2005 1/5/2006 1/5/2007 1/5/2008 1/5/2009 1/5/

60 Sector Plots Evolution of sector exposure for FTSE China 25 Index from 2005 to % 90% 80% 70% 60% 50% 40% 30% 20% Utilities Telecommunication Services Financials Consumer Staples Consumer Discretionary Industrials Materials Energy 10% 0% 1/5/2005 1/5/2006 1/5/2007 1/5/2008 1/5/2009 1/5/

61 Sector Plots Evolution of sector exposure for Nifty 50 Index from 2002 to % 90% 80% 70% 60% 50% 40% 30% 20% 10% Utilities Telecommunication Services Information Technology Financials Health Care Consumer Staples Consumer Discretionary Industrials Materials Energy 0% 61

62 Sector Plots Evolution of sector exposure for FTSE ASEAN Index from 2001 to % 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Utilities Telecommunication Services Information Technology Financials Health Care Consumer Staples Consumer Discretionary Industrials Materials Energy

63 Style Plots Evolution of style exposure for Hang SengIndex from 2002 to % 90% 80% 70% 60% 50% 40% 30% Growth Value 20% 10% 0% 63

64 Style Plots Evolution of style exposure for Nikkei 225 Index from 2002 to % 90% 80% 70% 60% 50% 40% 30% Growth Value 20% 10% 0% 64

65 Style Plots Evolution of style exposure for Topix 100 Index from 2002 to % 90% 80% 70% 60% 50% 40% 30% Growth Value 20% 10% 0% 65

66 Style Plots Evolution of style exposure for FTSE STI Index from 2002 to % 90% 80% 70% 60% 50% 40% 30% Growth Value 20% 10% 0% 66

67 Style Plots Evolution of style exposure for Kospi 200 Index from 2002 to % 90% 80% 70% 60% 50% 40% 30% Growth Value 20% 10% 0% 67

68 Style Plots Evolution of style exposure for FTSE TWSE 50 Index from 2002 to % 90% 80% 70% 60% 50% 40% 30% Growth Value 20% 10% 0% 68

69 Style Plots Evolution of style exposure for CSI 300 Index from 2003 to % 90% 80% 70% 60% 50% 40% 30% Growth Value 20% 10% 0%

70 Style Plots Evolution of style exposure for FTSE China 25 Index from 2002 to % 90% 80% 70% 60% 50% 40% Growth Value 30% 20% 10% 0% 70

71 Style Plots Evolution of style exposure for Nifty 50 Index from 2002 to % 90% 80% 70% 60% 50% 40% 30% Growth Value 20% 10% 0% 71