Nomura Global Quantitative Research Monthly 04-2012 510050

EQUITY RESEARCH

An index for long-short quants Style selection in emerging Asia Earnings surprise prediction

Global Quantitative Research Monthly

April 18, 2012

See Appendix A-1 for analyst certification, important disclosures and the status of non-US analysts.

Nomura | Global Quantitative Research Monthly April 18, 2012

2

Contents

Page

Chapter I A good start to the year for long-short quants 4

■ We add to our indices tracking the performance of quant funds with a new index for long-short quant strategies.

■ Long-short quant funds have had a good start to the year, delivering an 3.2% absolute return equal weighted (1.1% asset weighted).

■ We can divide the sample by fund and strategy type. Pure market neutral quant funds have delivered 0.7% return YTD equal weighted (1.2% asset weighted). We also show that dynamic beta allocation strategies have generally added value in recent years.

■ We compare our quant indices to established indices for the broader hedge fund community, the fit is generally close, though in 2012 quant funds have outperformed the broader market-neutral group.

Inigo Fraser-Jenkins - Global Head of Quantitative Strategy

Chapter II Style selection strategy in emerging Asia 11

■ The proportion of Asia emerging markets’ weightings has increased consistently in the region over the past decade. We take a look at the quant landscape and effectiveness of style picking strategy in emerging Asia.

■ We find that yield, valuation, and revision styles have consistently delivered a higher alpha in emerging Asia than that in developed Pacific ex-Japan.

■ Our style picking model has delivered better long-short performance and long-side alpha in EM Asia markets than their developed peers. This allows us to implement our model as a long-only strategy in EM Asia due to short-sell restrictions in some markets.

■ We show that the style model and alpha scores can be applied to the portfolio management process to practically implement the strategy in emerging Asia.

Sandy Lee - Head of Equity Quantitative Strategies, Asia ex-Japan

Chapter IIIEarnings surprise forecasting strategy 27

■ The 12/3 full-year results announcement season will soon be upon us■ We attempt to predict which stocks will produce an earnings surprise, mainly on the basis of the continuity of

individual stocks' surprise bias.Akihiro Murakami - Chief Quantitative Strategist, Japan


3

Foreword We extend our tracking of quant fund performance to include global long-short quant funds. These funds have shown generally good performance this year. Moreover, we hope that this index will be a useful benchmark for long-short approaches.

We introduce a style selection strategy for emerging Asian markets, where we see strong potential for style allocation. We also discuss a model for predicting earnings surprise in Japan.

Our annual quant conferences in London and New York will be held in coming months. We hope these will give investors an opportunity for a discussion of the key current issues in quant investment management. Details are below; we look forward to seeing you there.

Nomura Annual Quantitative Conferences in London and New York:

Nomura Global Quantitative Equity Conference to be held at our offices in London on Tuesday 22 May 2012.

Our Global Quantitative Equity Conference in London, as in previous years, includes external and internal speakers. External speakers will include Professor Narasimhan Jegadeesh, Dean’s Distinguished Chair of Finance at Emory University, who will discuss a new approach for content analysis and other recent research and a number of buy side speakers including Daniele Scognamiglio, Fund Manager Quant Equity & Gerben de Zwart, Head of Quantitative Research from APG Asset Management who will talk about Cross-industry news sentiment. The Nomura speakers will be the heads of quant research in each region. We hope that you will join us along with your fellow leaders in the field of quantitative investment at this special event for a stimulating session of learning and discussion. For further information and to register for the conference in London please go to web page: https://www.nomura.com/events/global-quantitative-equity-conference/

Nomura Sixth Annual Global Quantitative Investment Strategies Conference to be held at The Pierre Hotel in New York on Tuesday 26 June 2012

For further information and to register for the conference in New York please contact: Nomura Quantitative Advisory Nomura Securities International, Inc. Telephone: 212 667 2081 Email: [email protected]


4

A good start to the year for long-short quants • We add to our indices, tracking the performance of quant funds with a new index for

long-short quant strategies.

• Long-short quant funds have had a good start to the year, delivering a 3.2% absolute return equal weighted (1.1% asset weighted).

• We can divide the sample by fund and strategy type. Pure market neutral quant funds have delivered 0.7% return YTD equal weighted (1.2% asset weighted). We also show that dynamic beta allocation strategies have generally added value in recent years.

• We compare our quant indices to established indices for the broader hedge fund community; the fit is generally close, though in 2012 quant funds have outperformed the broader market-neutral group.

We have been tracking the performance and asset flows of long-only equity quants for several years; we have now expanded this to also track long-short quant managers. 2012 has been a good year for long-short quants this year with an absolute return of 3.12%; this return makes up for lost ground since Q3 2011. Performance has been weaker since March, but to a large extent this reflects the presence of funds with a non-zero beta exposure in the sample.

Fig. 1: Performance of long-short quant managers

Source: Bloomberg, Nomura Quantitative research

There is a legitimate question about what the appropriate peer benchmark is for a quant manager. Should it just be a measure of the performance of active managers in general, or are there issues which are peculiar to quants? Moreover, over the last five years the word ‘quant’ has been applied to an increasingly broad church. Should different categories of quant approach even be compared with each other?

The reason we developed our quant long-only index was that there did not appear to be a good benchmark for long-only quants, and to our knowledge nothing that could capture the performance of different subcategories of quant. As we discuss below, one reason for this is that it is not always easy to spot a quant fund; many do not use the word quant in their description. Some have even explicitly gone into hiding in recent years and present a very non-quant façade to the world. This makes it hard for many index providers to construct a representative index. We have also been able to break out sub-indices by fund type and strategy to provide a more granular set of indices. With this report we are attempting to construct an equivalent set of indices for long-short quant managers. In the case of long-short indices there are, in some sense, more readily available benchmarks. For example, the Hedge Fund Research Equity Market Neutral index is often taken as a proxy for long-short quants as quant approaches form a large

80

90

100

110

120

130

140

150

160

Jan

-06

Ap

r-0

6

Jul-0

6

Oct

-06

Jan

-07

Ap

r-0

7

Jul-0

7

Oct

-07

Jan

-08

Ap

r-0

8

Jul-0

8

Oct

-08

Jan

-09

Ap

r-0

9

Jul-0

9

Oct

-09

Jan

-10

Ap

r-1

0

Jul-1

0

Oct

-10

Jan

-11

Ap

r-11

Jul-1

1

Oct

-11

Jan

-12

Ap

r-1

2

Index

Qhole sample - (eq wgt)

Whole sample - (asset wgt)

Inigo Fraser Jenkins

+44 207 102 4658

[email protected]

Gerard Alix Guerrini

+44 207 102 5079

[email protected]

Quant long-short funds have performed well so far this year

mailto:[email protected]�



5

part of the index. We discuss the relationship between our index and third party indices below. Our index differs as it is the only one (of which we are aware) that explicitly focuses on quant funds. We also break out various sub indices in terms of market exposure and strategy type.

As with our analysis of long-only managers, building up a list of long-short quant managers is non-trivial. We adopted a two-stage process. The first was to screen for funds that describe themselves as quant or systematic. The second stage was to use our own proprietary list of managers whom we know to run long-short quant strategies and to identify data on the performance of their funds. Our objective was to keep some homogeneity in the sample so to that end we include only funds that are primarily quant-driven and also limit the sample to funds that are equity-only. This yields a sample of 69 funds. We are very alive to survivorship bias issues in constructing such an index, so we include dead funds where we can find data. We recognise that it is difficult to capture all dead funds, so survivorship bias becomes more challenging as we push the index further back through time. Our sample includes six such funds at present. We intend for this sample to grow over time.

Analysing long-short quant performance is more complicated than for long-only because of the even greater differences between managers. There is a wide spread of market exposures and of strategies.

To discriminate between the several levels of market exposure in the long-short universe, we calculated the betas relative to their regional ‘long-only’ benchmarks; the returns are defined on a monthly basis as the lowest publication frequency in our sample is monthly. Indeed, if we look at Figure 2 below, we can easily infer three clusters of funds regarding their averaged beta 12mth values. For the beta interval between -0.1 and +0.2 we named the subset as ‘Pure Market Neutral’, then ‘Intermediary’ for the subset between 0.2 and 0.6 to end with the bigger than 0.6 beta group that can be described as ‘Not Market Neutral’. We break out performance data for the overall sample and also for these beta clusters separately.

Fig. 2: Distribution of market betas for long-short sample


0

1

2

3

4

5

6

-0.0

75

-0.0

25

0.0

25

0.0

75

0.1

25

0.1

75

0.2

25

0.2

75

0.3

25

0.3

75

0.4

25

0.4

75

0.5

25

0.5

75

0.6

25

0.6

75

0.7

25

0.7

75

0.8

25

0.8

75

0.9

25

0.9

75

1.0

25

Fre

que

ncy

Beta

There are difficulties in constructing an index for quant managers

There are various clusters of beta exposure in the fund sample


6

When any new index is presented, it is natural to try to compare it to similar attempts that have been made before. We think that two benchmarks are suitable for comparison purposes: The Hedge Fund Research Equity Market Neutral Index and the Eurekahedge Long Short Equities Hedge Fund Index. Both are widely recognised, though they have slightly different properties.

In comparing our fund sample to the two different benchmarks, it makes sense for us to split our long-short quant index into smaller subsets. These groups are distinguished by their beta relative to their regional ‘long-only’ benchmarks. The subset that delivers the closest to zero beta is described as ‘pure market neutral index’

We compare our overall long-short quant equity index to the Eurekahedge Long Short Equities Hedge Fund Index as this index represents the wide equity long-short universe, which is not strictly market neutral. The methodology of the equally or asset weighted indices from Eurekahedge is very close to our own as these indices keep the dead funds and aggregate new funds when they match their criteria, which implies a possible impact of the historical returns. However, our universe is smaller as we focus on ‘quant funds’; the only difference is their average of the fund returns in the local currency compared with our homogeneous dollar-denominated returns.

We compare the two indices below in Figure 3. They both have a net market exposure, which is evident in their return profile over the last few years, but the two have moved closely together.

Fig. 3: Long-short quant index and Eurekahedge Long Short Equities Hedge Fund Index


We compare our ‘pure market neutral index’ to the Hedge Fund Research Equity Market Neutral Index. This index, which is built by clustering and simulations, aims to detect the element of the hedge fund universe that uses market neutral strategies. They screen the funds to check if they are correlated to the market in order to decide whether to incorporate them or not. Although this index is built with a different methodology than ours, the core idea remains the same, albeit of course the fact that our universe is smaller as we focus on equity quantitative long-short strategies and so we think the underlying funds selected by the index have a high likelihood to overlap.

We compare these two indices in Figure 4. Here there are greater differences between the indices; however, they have responded to some key events in a very similar way. The ‘quant event’ of August 2007 is a prominent feature of both indices shown as a sharp sell-off and recovery. They also both showed good performance in mid 2008, followed by a sell-off. In 2011 both did well. However, what is intriguing is that our pure market neutral index has shown a good performance so far in 2012, whereas the Hedge Fund Research Index has shown weaker performance.

90

100

110

120

130

140

150

160

De

c-0

5F

eb

-06

Ap

r-0

6Ju

n-0

6A

ug

-06

Oct

-06

De

c-0

6F

eb

-07

Ap

r-0

7Ju

n-0

7A

ug

-07

Oct

-07

De

c-0

7F

eb

-08

Ap

r-0

8Ju

n-0

8A

ug

-08

Oct

-08

De

c-0

8F

eb

-09

Ap

r-0

9Ju

n-0

9A

ug

-09

Oct

-09

De

c-0

9F

eb

-10

Ap

r-1

0Ju

n-1

0A

ug

-10

Oct

-10

De

c-1

0F

eb

-11

Ap

r-11

Jun

-11

Au

g-1

1O

ct-1

1D

ec-

11F

eb

-12

Index

Eureka L/S HF Index

Whole sample


7

Fig. 4: Pure Market Neutral quant index and Hedge Fund Research Equity Market Neutral Index


The AUM of our fund sample has been relatively static at about USD 7bn over the past three years, see Figure 5.

Fig. 5: AUM of fund sample

Source: Nomura Quantitative research

Within the group, 30 funds are global;14 cover North America, 17 are European focussed and 8 are Asian, Figure 6.

Fig. 6: Distribution of funds by region

Source: Nomura Quantitative research

We can subdivide the overall group by region and by market exposure. We show this both for the overall sample and for the subset that is pure market neutral (ie with a beta close to zero).

80

85

90

95

100

105

110

115

120

125

130

Jan

-06

Ma

r-0

6M

ay-

06

Jul-0

6S

ep

-06

No

v-0

6Ja

n-0

7M

ar-

07

Ma

y-0

7Ju

l-07

Se

p-0

7N

ov-

07

Jan

-08

Ma

r-0

8M

ay-

08

Jul-0

8S

ep

-08

No

v-0

8Ja

n-0

9M

ar-

09

Ma

y-0

9Ju

l-09

Se

p-0

9N

ov-

09

Jan

-10

Ma

r-1

0M

ay-

10

Jul-1

0S

ep

-10

No

v-1

0Ja

n-1

1M

ar-

11M

ay-

11Ju

l-11

Se

p-1

1N

ov-

11Ja

n-1

2M

ar-

12

Index

HFRXEMN Index

Pure Market Neutral

0

1000

2000

3000

4000

5000

6000

7000

8000

Jan

-06

Ap

r-0

6

Jul-0

6

Oct

-06

Jan

-07

Ap

r-0

7

Jul-0

7

Oct

-07

Jan

-08

Ap

r-0

8

Jul-0

8

Oct

-08

Jan

-09

Ap

r-0

9

Jul-0

9

Oct

-09

Jan

-10

Ap

r-1

0

Jul-1

0

Oct

-10

Jan

-11

Ap

r-11

Jul-1

1

Oct

-11

Jan

-12

Ap

r-1

2

Index

Region Number of funds

Global 30North America 14Europe 17Asia 8

The funds are broadly diversified by region


8

Fig. 7: Performance of quant long-short funds by region


Fig. 8: Performance of quant long-short funds by region – pure market neutral subset


In addition to dividing the sample by beta and region, we also can subdivide it by strategy type. We have done this previously for our sample of long-only funds and the feedback was that it offered an insight into the relative attractiveness of strategies or fund types that was hard to achieve otherwise. Our technique for doing this is through a survey and asking all the managers in the sample to self-classify their funds by a number of metrics. There are always issues with surveys and withy self-classification, but we feel that it is useful when there is no other practical way to obtain the information. We would like to thank all the managers who agreed to take part.

90

100

110

120

130

140

150

160

170Ja

n-0

6

Ap

r-0

6

Jul-0

6

Oct

-06

Jan

-07

Ap

r-0

7

Jul-0

7

Oct

-07

Jan

-08

Ap

r-0

8

Jul-0

8

Oct

-08

Jan

-09

Ap

r-0

9

Jul-0

9

Oct

-09

Jan

-10

Ap

r-1

0

Jul-1

0

Oct

-10

Jan

-11

Ap

r-11

Jul-1

1

Oct

-11

Jan

-12

Ap

r-1

2

Index

Whole sample - (eq wgt)Global fundsNorth AmericaEuropeAsiaWhole sample - (asset wgt)

80

90

100

110

120

130

140

150

160

Jan

-06

Ap

r-0

6

Jul-0

6

Oct

-06

Jan

-07

Ap

r-0

7

Jul-0

7

Oct

-07

Jan

-08

Ap

r-0

8

Jul-0

8

Oct

-08

Jan

-09

Ap

r-0

9

Jul-0

9

Oct

-09

Jan

-10

Ap

r-1

0

Jul-1

0

Oct

-10

Jan

-11

Ap

r-11

Jul-1

1

Oct

-11

Jan

-12

Ap

r-1

2

Index

Global funds

North America

Asia


9

We asked managers to classify their strategy along a number of parameters; the results are shown in Figure 9. The investment process is mainly split between pure quant approaches and quant with broad discretionary inputs. A quarter of the funds (by number) use a quant pre-screening but then use other approaches for final stock selection. The vast majority of the funds use a dynamic factor allocation process. This is in-line with the findings of our work on long-only quant funds where dynamic strategies have overtaken static approaches over the past five years.

Interestingly, the majority of the funds employ strategies where the factors are imposed on the model, rather than being derived from an empirical (eg regression) process. We discuss the performance implications of this below.

Nearly all the funds employ a ‘rebalance anytime’ approach. This sets them apart from their long-only brethren where such a flexible approach only makes up about a third of the sample. The evidence is that the flexible rebalance approach strongly outperforms and has lower volatility, so this preference for flexibility among long-short funds is not surprising and eminently sensible.

Fig. 9: Description of long-short fund sample


If we look at the performance of these subgroups, some patterns emerge. Funds that allow for a dynamic adjustment of their market beta have fared better in absolute terms than funds where the beta is static, Figure 10. The choice between dynamic or static factor allocation does not adjust performance within the sample, though the dynamic factor allocation group does have greater variation in performance over the cycle.

Proportion ofFunds, %

Proportion of AUM, %

Investment processPure quant/systematic 42% 20%Quant with discretionary views on single stocks 0% 0%Quant with broad discretionary views possible 33% 35%Quant pre screening 25% 45%

Choice of factors and choice of weightsStatic 33% 32%Dynamic 78% 68%

Choice of factors and choice of weightsDerived f rom an empirical model (eg historic regression) 44% 7%Imposed on the model (eg an a priori decision on what to use) 67% 94%

Rebalance processWeekly 11% 0%Monthly 11% 5%Quarterly 0% 0%At any time 78% 95%

FactorsTechnical/momentum 55% 52%Accounting data-based 91% 97%Macro 36% 32%Company-specif ic but not accounts-based 45% 50%Derivatives-based 27% 31%

Target market exposureStatic 56% 94%Dynamic 44% 6%


10

Fig. 10: Dynamic beta allocation outperforms static allocation


There is a large difference in performance between funds where the factors are selected and weighted as the result of an inductive process as opposed to a deductive process. That is to say, an empirical process, such as regressions to decide what factors to use and how to weight them, strongly outperforms an approach where the factors are imposed by the manager; this result differs from the long-only world where the results of the two approaches are closer. The result is also interesting in light of the AUM of the deductive strategies being greater than for the inductive strategies.

Fig. 11: Inductive models outperform deductive models


We hope that this index will be useful for long-short quant managers as a guide to the performance of the industry. We plan to monitor it on a regular basis; we also will grow it over time. This should allow us to report more granular results on the performance of subsections of the index, the reporting of which is limited if the sample of any given subcategory is too small.

50

70

90

110

130

150

170

190

210

230D

ec-

07

Fe

b-0

8

Ap

r-0

8

Jun

-08

Au

g-0

8

Oct

-08

De

c-0

8

Fe

b-0

9

Ap

r-0

9

Jun

-09

Au

g-0

9

Oct

-09

De

c-0

9

Fe

b-1

0

Ap

r-1

0

Jun

-10

Au

g-1

0

Oct

-10

De

c-1

0

Fe

b-1

1

Ap

r-11

Jun

-11

Au

g-1

1

Oct

-11

De

c-11

Fe

b-1

2

Ap

r-1

2

Index

50

70

90

110

130

150

170

190

210

De

c-0

7

Fe

b-0

8

Ap

r-0

8

Jun

-08

Au

g-0

8

Oct

-08

De

c-0

8

Fe

b-0

9

Ap

r-0

9

Jun

-09

Au

g-0

9

Oct

-09

De

c-0

9

Fe

b-1

0

Ap

r-1

0

Jun

-10

Au

g-1

0

Oct

-10

De

c-1

0

Fe

b-1

1

Ap

r-11

Jun

-11

Au

g-1

1

Oct

-11

De

c-11

Fe

b-1

2

Ap

r-1

2

Index


11

Style selection strategy in emerging Asia

• The proportion of Asia emerging markets’ weightings has increased consistently in the region over the past decade. We take a look at the quant landscape and effectiveness of style picking strategy in emerging Asia.

• We find that yield, valuation, and revision styles have consistently delivered a higher alpha in emerging Asia than that in developed Pacific ex-Japan.

• Our style picking model has delivered better long-short performance and long-side alpha in EM Asia markets than their developed peers. This allows us to implement our model as a long-only strategy in EM Asia due to short-sell restrictions in some markets.

• We show that the style model and alpha scores can be applied to the portfolio management process to practically implement the strategy in emerging Asia.

Growing interest in emerging Asia

Over the past ten years, there has been a growing impact from Asia on the rest of the world, and Asian markets have gained weightings steadily in the global equity markets. The chart below reveals that Asia emerging markets, in particular, have captured market weight at a faster pace than their developed peers, reflecting faster development in economies and a growing investment preference in the region.

Fig. 12: Weightings of emerging Asia and developed Pacific ex Japan in MSCI AC World

Note: Universe for calculating weightings is based on MSCI AC Asia Pacific ex Japan.

Source: MSCI, Nomura Quantitative Strategies

The chart below (left) exhibits the breakdown of investible weightings by developed and emerging markets in Asia Pacific ex Japan region. The proportion of Asia emerging markets’ weightings has increased consistently over the past decade, now accounting for roughly 61% of the regional total. The weightings of China, India, Indonesia, Korea, the Philippines and Thailand gain on a relative basis. By country, China is the biggest weighting gainer, with its market weighting in Asia Pacific ex Japan rising to 17.9% at end-March 2012 from just 5.4% at end-2001, as shown below (right)

0

1

2

3

4

5

6

7

8

9

20

01

20

02

20

03

20

04

20

05

20

06

20

07

20

08

20

09

20

10

20

11

20

12

We

igh

ting

s %

of W

orl

d

Developed Pacific ex-Japan

Emerging Asia

Asia emerging markets have captured market weight at a faster pace than their developed peers

Sandy Lee

+852 2252 2101

[email protected]

Rico Kwan

+852 2252 2102

[email protected]

Yasuhiro Shimizu

+852 2252 2107

[email protected]

Previously published on 13 April 2012

Weightings of China, India, Indonesia, Korea, the Philippines, and Thailand gain on a relative basis





12

Fig. 13: Weightings of developed and emerging markets in Asia Pacific ex Japan

Note: Universe for calculating weightings is based on MSCI AC Asia Pacific ex Japan. Source: MSCI, Nomura Quantitative Strategies

Fig. 14: Asia Pacific ex Japan’s weightings by country

Note: Weightings are as at each calendar year end. 2012 figures are as of end March 2012. AU: Australia, SG: Singapore, HK: Hong Kong, CN: China, KR: Korea, TW: Taiwan, IN: India. Source: MSCI, Nomura Quantitative Strategies

Judging from a survey of participants in Nomura Global Quantitative conferences held in European cities in May 2011 and general feedback from client meetings in recent years, Emerging Market Quant is seen as one of the areas where global fund management community can win assets. The chart below (left) shows that the percentage of emerging market’s AUM share in global equity mutual funds dropped in 2011, but has increased so far in 2012. Performance-wise, emerging Asia stock markets underperformed their developed peers in H2 2011, but have outperformed in 2012 YTD.

Fig. 15: AUM share of emerging market equities mutual funds

Note: Data run to 11 April. Source: EPFR Global, Nomura Quantitative Strategies

Fig. 16: Relative performance

Source: Thomson Reuters Datastream, Nomura Quantitative Strategies

We take a look at the recent flow of funds tracking equity mutual fund net subscriptions. Year-to-date global emerging equity markets have attracted inflows, while the developed markets have seen net outflows. Global equity mutual fund investors invested a net USD21.9bn into all dedicated emerging market funds, recording positive net inflows in 13 out of 15 weeks since the beginning of 2012. In Asia Pacific ex-Japan, YTD mutual fund investors invested a net USD131mn into Asia equity funds. Total net injections into emerging Asia equity markets were USD352mn versus net outflows of USD221mn from developed Pacific ex Japan markets. The net inflows into emerging Asia markets that we have seen YTD undo just around 2.4% of the mutual fund net outflows in 2011

30

35

40

45

50

55

60

65

70 2

00

1

20

02

20

03

20

04

20

05

20

06

20

07

20

08

20

09

20

10

20

11

20

12

We

igh

ting

s %

of A

PxJ

Developed Pacific ex-Japan

Emerging Asia

0%10%20%30%40%50%60%70%80%90%100%

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

AU SG HK CN KR TW IN Others

10%

12%

14%

16%

18%

20%

De

c-0

9

Ma

r-1

0

Jun

-10

Se

p-1

0

De

c-1

0

Ma

r-1

1

Jun

-11

Se

p-1

1

De

c-1

1

Ma

r-1

2

% of EM in global equity mutual funds

% of EM in global equity mutual funds (ex-ETFs)

90

95

100

105

110

115

De

c-0

9

Ma

r-1

0

Jun

-10

Se

p-1

0

De

c-1

0

Ma

r-1

1

Jun

-11

Se

p-1

1

De

c-1

1

Ma

r-1

2

31 Dec 2009 = 100

Emerging Asia vs developed Pacific ex Japan

Emerging Market Quant is seen as one of the areas where the global fund management community can win assets

YTD global emerging equity markets attracted more inflows than the developed markets


13

Fig. 17: Net equity flows (Global EM versus DM)


Fig. 18: Net equity flows (regional EM versus DM)


Given growing investors’ interest in emerging Asia, we thought it would be useful to give an overview of the general landscape and style selection strategy in Asia emerging markets through the following analysis. Hopefully this will help take stock of the key issues in investing in the emerging Asia region and provide some colour on the merits of the style selection model in the region.

Return attribution: Rising global factor; country effect still dominates

Understanding the different factors that drive stock returns is important in investing in emerging Asia markets. We update a return attribution analysis for Asia Pacific equities, using same methodology in Quantitative Landscape (Shimizu, 18 August 2010). We estimate changes in the stock returns attributable to four factors: the global market, country-, sector- and stock-specific factors. The implications for investors are obvious, since a better understanding of these factors should help provide valuable information in formulating equity allocation strategies.

We assume that the return of the stock i can be explained by this multifactor model.

itktIijt

Cit

Miiit ICMR

where Rit = the return of stock i denominated in local currency at time t

Mt, = the return of the FTSE All-World Index

Cjt and Ikt = the country and global industry factor return at time t, respectively

Appendix I gives further details on the methodology and how we estimate each coefficient. We calculate the importance of the global, country and sector factor effect in explaining variations in stock return R using βM, βC, and βI, respectively.

it

tMi

R

MfactorGlobal

var

var　　　

it

jtCi

R

CfactorCountry

var

var　　　

it

ktIi

R

IfactorSector

var

var　　　　

it

it

RfactorstockIndividual

var

var　　　　

-60000

-40000

-20000

0

20000

40000

60000

80000

100000Ja

n-1

0

Ap

r-1

0

Jul-

10

Oct

-10

Jan

-11

Ap

r-1

1

Jul-

11

Oct

-11

Jan

-12

Ap

r-1

2

USDmn All dedicated EM equities fundsAll dedicated DM equities funds

-5000

0

5000

10000

15000

20000

Jan

-10

Ap

r-1

0

Jul-

10

Oct

-10

Jan

-11

Ap

r-1

1

Jul-

11

Oct

-11

Jan

-12

Ap

r-1

2

USDmn EM Asia equities funds

Pacific ex Japan equities funds

We analyse the return attributions from global market, country-, sector- and stock-specific factors


14

Rising trend of the global factor effect

The chart below shows the changes in the global factor effect from 1994 to 2011 for Japan, developed Pacific ex-Japan, and emerging Asia.

Fig. 19: Global factor effect (1994 to 2011)

Source: FTSE, Nomura Quantitative Strategies

In the aftermath of the 2008 credit crisis, we observed a rising global factor effect in Asian markets amid a rise of global asset correlation. In the region, both developed and emerging markets have become more sensitive to the global market. The rising trend has continued over recent years, especially in emerging Asia.

The chart below presents the global factor effect in 2011 for each of the Asia Pacific markets. Despite a rising trend of global factor effect, emerging Asia markets such as Indonesia, India, Thailand, Malaysia, Philippines and Pakistan had the lowest global factor effect, whereas developed markets such as Australia and Singapore saw the opposite. This gives investors a means to diversify their global portfolio. Among Asia emerging markets, the global factor effect was most important in Taiwan and Korea.

Fig. 20: Global factor effect for Asia Pacific markets (2011)

Note: AUS: Australia, SGP: Singapore, TWN: Taiwan, KOR: Korea, CHN: China, HKG: Hong Kong, JPN: Japan, IDN: Indonesia, IND: India, THA: Thailand, MYS: Malaysia, PHL: Philippines, NZL: New Zealand, PAK: Pakistan.


0

5

10

15

20

25

30

35

40

45

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

JPN Developed ex JPN Emerging Asia

(%)

0

5

10

15

20

25

30

35

40

AUS SGP TWN KOR CHN HKG JPN IDN IND THA MYS PHL NZL PAK

Global factor

(%)

We see a rising global factor effect in Asia

We see a rising global factor effect in Asia; by market, the global factor effect is the strongest in developed markets such as Australia and Singapore


15

Country and sector factor effects

The chart below shows the spread between the country factor and sector factor effects in Japan, developed Pacific ex-Japan and emerging Asia. Overall on average, the country factor effect is stronger than the sector factor effect in the Asia Pacific ex Japan region. In the developed markets, the difference between the country factor and sector factor effects is relatively small. In contrast, our return attribution analysis shows that the country effect still dominates the industry effect in emerging Asia. However, the spread is getting narrower, signaling a converging characteristic of emerging Asia to the developed markets.

Fig. 21: Spread between country and sector factor effects (1994 to 2011)


The chart below reveals the country and sector effect in each of the Asia Pacific markets. The country factor effect is generally stronger than the sector effect in most markets in the region. But there were some clear differences in the country and sector factor effect among markets in Asia emerging and in developed Pacific ex Japan.

Fig. 22: Country and sector factor effect (2011)


0

5

10

15

20

25

30

35

40

45

50

PAK PHL IDN THA HKG CHN IND SGP JPN TWN MYS KOR AUS NZL

Industry factor

Country factor

(%)

The country effect still dominates the industry effect in emerging Asia, but the spread is getting narrower

The country factor effect is generally stronger than the sector effect in most markets in the region


16

Among developed Asia markets, the country effect is relatively more important in Hong Kong, but not as strong in Australia and Japan. Among Asia emerging markets, the industry factor effect is more important in Taiwan and Korea. In contrast, the country effect is most significant in Pakistan, the Philippines, Indonesia, Thailand and China.

In summary, the results of the return attribution analysis suggest a rising global factor effect in Asia. On average, the country effect dominates the industry effect as an explanation of the sources of stock return variation especially in emerging Asia, although the spread is getting narrower. Despite the increasing globalization of emerging Asia markets, country diversification seems to be more efficient than diversification across industries

Performance of investment styles

Despite closer tracking of emerging Asian stocks than ever in recent years by global investors, this doesn’t stop them from delivering a better return than developed markets in terms of quantitative factor alpha. We examine the performance of eight investment style categories, namely, size, momentum, yield, valuation, revision, growth, profitability, and risk, using the same methodology and approach described in Style Selection Model (15 March 2011). We investigate the style returns in the emerging Asia region and the developed Pacific ex-Japan region, respectively.

Fig. 23: Definition of eight investment styles

Style Factors Definition

Size * Market cap Log of US$ market cap

Momentum Price momentum (12M -1M)

Last 12-month return less the last 1-month return in local currency

Yield Dividend yield F12-month DPS / stock price

Valuation Earnings yield F12-month EPS / stock price

B/P Actual BPS / stock price

Revision Revision index (Number of upward analyst revisions - number of downward analyst revisions) / total number of analysts' estimate

Growth Sales growth (FY2) FY2 sales / FY1 sales

EPS growth (FY2) FY2 EPS / FY1 EPS

Profitability Return on equity F12-month net profit / actual shareholders' equity

Change in pre-tax profit margin

FY2 pre-tax profit margin - FY1 pre-tax profit margin

Risk * Volatility Past 36-month price return volatility

Estimate dispersion I/B/ES FY1 consensus EPS standard deviation / absolute value for FY1 consensus EPS

Default probability Default probability estimated using Merton model

Note: Style scores are calculated by summing the normalized scores of the underlying factors. The styles marked with * are reverse-based. Reported data are sourced from Worldscope, consensus estimate data from I/B/E/S, and price data from Exshare.

Source: Nomura Quantitative Strategies

Country effect is most significant in emerging Asia markets such as Pakistan, the Philippines, Indonesia, Thailand, and China

We examine the performance of eight investment styles


17

Emerging Asia saw a higher alpha in general

We note that there were clear differences between emerging Asia and developed Pacific ex-Japan markets in the performance of particular styles. Over the long run, yield, valuation and revision styles have consistently delivered a higher alpha in emerging Asia than that in developed Pacific ex-Japan. This was even more obvious after the credit crisis, where both value investing and yield have been the two most promising styles in the region since 2009, delivering a significant return of 16.0% and 7.2% per annum respectively in emerging Asia, whereas only rising 5.0% and 0.6% per annum respectively in developed Pacific ex-Japan.

In contrast, we find that there was a convergence in the performances of profitability and risk measures between emerging Asia and developed Pacific ex-Japan, especially after the credit crisis. Before 2009, barely investing in high profitability companies did not generate significant alpha in emerging Asia, whereas this style worked well in developed Pacific ex-Japan between 2004 and 2008. Since 2009, the profitability style has worked in-line in both the emerging Asia and developed Pacific ex-Japan regions, suggesting investors are now focusing more on the operating quality of emerging Asia companies than before, and weight this measure similarly as that in developed Pacific ex-Japan.

From a risk perspective, we find investors were more risk averse in emerging Asia than that in developed Pacific ex-Japan before the credit crisis, with low-risk stocks being rewarded in emerging Asia but high-risk stocks outperforming their low-risk peers in developed Pacific ex-Japan. However, since 2009, the performance of risk style has been aligned with the developed and emerging Asia regions.

Yield, valuation and revision styles have consistently delivered a higher alpha in emerging Asia than that in developed Pacific ex-Japan in the long run

Since the credit crisis, we have seen a convergence in the performances of profitability and risk measures between emerging Asia and developed Pacific ex-Japan


18

Fig. 24: Performance of eight investment styles in developed Pacific ex-Japan and emerging Asia

Note: The styles marked with * are reverse-based. Source: Worldscope, I/B/E/S, MSCI, Nomura Quantitative Strategies

Size * Momentum

Yield Valuation

Revision Growth

Profitability

Risk *

-30-20-10

0102030405060

Dec

-98

Dec

-99

Dec

-00

Dec

-01

Dec

-02

Dec

-03

Dec

-04

Dec

-05

Dec

-06

Dec

-07

Dec

-08

Dec

-09

Dec

-10

Dec

-11

(%)

Developed Pacific ex-JP

Emerging Asia-40

-20

0

20

40

60

80

100

120

Dec

-98

Dec

-99

Dec

-00

Dec

-01

Dec

-02

Dec

-03

Dec

-04

Dec

-05

Dec

-06

Dec

-07

Dec

-08

Dec

-09

Dec

-10

Dec

-11

(%)


Emerging Asia

-40

-20

0

20

40

60

80

100

Dec

-98

Dec

-99

Dec

-00

Dec

-01

Dec

-02

Dec

-03

Dec

-04

Dec

-05

Dec

-06

Dec

-07

Dec

-08

Dec

-09

Dec

-10

Dec

-11

(%)


Emerging Asia

-50

0

50

100

150

200

Dec

-98

Dec

-99

Dec

-00

Dec

-01

Dec

-02

Dec

-03

Dec

-04

Dec

-05

Dec

-06

Dec

-07

Dec

-08

Dec

-09

Dec

-10

Dec

-11

(%)


Emerging Asia

-200

20406080

100120140160

Dec

-98

Dec

-99

Dec

-00

Dec

-01

Dec

-02

Dec

-03

Dec

-04

Dec

-05

Dec

-06

Dec

-07

Dec

-08

Dec

-09

Dec

-10

Dec

-11

(%)


Emerging Asia

-5

0

5

10

15

20

25

30

Dec

-98

Dec

-99

Dec

-00

Dec

-01

Dec

-02

Dec

-03

Dec

-04

Dec

-05

Dec

-06

Dec

-07

Dec

-08

Dec

-09

Dec

-10

Dec

-11

(%)


Emerging Asia

-20

-10

0

10

20

30

40

50

Dec

-98

Dec

-99

Dec

-00

Dec

-01

Dec

-02

Dec

-03

Dec

-04

Dec

-05

Dec

-06

Dec

-07

Dec

-08

Dec

-09

Dec

-10

Dec

-11

(%)


Emerging Asia

-40

-30

-20

-10

0

10

20

30

40

Dec

-98

Dec

-99

Dec

-00

Dec

-01

Dec

-02

Dec

-03

Dec

-04

Dec

-05

Dec

-06

Dec

-07

Dec

-08

Dec

-09

Dec

-10

Dec

-11

(%)Developed Pacific ex-JP

Emerging Asia


19

Turnover of style portfolios

The charts below compare the average one-way turnover in the long sides of style portfolios in developed Pacific ex-Japan and in emerging Asia, over the long run and over the past one year. We find that the turnover rate of style portfolios has been generally higher in emerging Asia than in developed Pacific ex-Japan.

Fig. 25: Average one-way turnover of style portfolios

Source: Worldscope, I/B/E/S, MSCI, Nomura Quantitative Strategies

Low correlations in style returns

We also examine the average pair-wise correlation between the returns of eight investment styles in developed Pacific ex-Japan and in emerging Asia. The style correlations are generally low (or negative) in both sub-regions, reflecting that these styles offer diversification opportunities. Notably, such low correlation appears to be more consistent in emerging Asia than that in developed Pacific ex-Japan. Since late 2010, the style correlation has been lower (more negative) in emerging Asia than that in developed Pacific ex-Japan.

Fig. 26: Average correlation of quant styles

Note: Chart shows 24-month rolling average pair-wise correlation of return for the eight investment styles. Style performance is calculated by cumulating the return spread between top quintile and bottom quintile based on a composite style score. Style portfolios are rebalanced monthly with country and sector diversification.


0

10

20

30

40

50

60

70

Siz

e

Mo

men

tum

Yie

ld

Valu

atio

n

Revis

ion

Gro

wth

Pro

fita

bili

t y

Ris

k

1-w

ay lo

ng tu

rnove

r (%

)

Long run since 1999


Emerging Asia

0

10

20

30

40

50

60

70

Siz

e

Mo

men

tum

Yie

ld

Val

uatio

n

Rev

isio

n

Gro

wth

Pro

fitab

ilit y

Ris

k

1-w

ay lo

ng

turn

over

(%)

Past one year


Emerging Asia

-0.12

-0.10

-0.08

-0.06

-0.04

-0.02

0.00

0.02

0.04

0.06

0.08

Dec-0

0

Dec-0

1

Dec-0

2

Dec-0

3

Dec-0

4

Dec-0

5

Dec-0

6

Dec-0

7

Dec-0

8

Dec-0

9

Dec-1

0

Dec-1

1

Correlation


Emerging Asia

The turnover rate of style portfolios has been generally higher in emerging Asia than in developed Pacific ex-Japan

Since late 2010, the style correlation has been lower in emerging Asia than that in developed Pacific ex-Japan


20

Style selection model in emerging Asia

In the previous section, we discussed that the magnitudes of alphas extracted are generally larger in emerging Asia than those in developed Asia, alongside a low correlation in style returns, suggesting that there are opportunities for the outperformance of the style-picking model in this sub-region. In this section, we attempt to run our existing style selection model (Style Selection Model, 15 March 2011) in emerging Asia.

The chart below illustrates the style ranking process of our style selection model. We select the two styles with the best ranking scores in a weighted average approach based on the long-term relative value spread and short-term relative strength of the style, plus adding on a risk-aversion proxy.

Fig. 27: Illustration of the style selection model ranking process


We then compare the return differences of the style selection model between emerging Asia and develop Pacific ex-Japan, and study the opportunities of extracting the long side alpha in emerging Asia as practical factors such as stock lending costs and availability may make it difficult to short-sell in some markets. Our objective is to realistically implement the style-picking strategy for optimal performance in emerging Asia.

Performance of style selection models

The chart below presents the cumulative long-short performances of our style selection model in developed Pacific ex-Japan and emerging Asia using average return of the selected styles. Since December 2000, the model has delivered a long-short return of 12.2% pa with an information ratio (IR) of 2.1 in emerging Asia, outperforming the long-short return of 6.6% pa (IR as 1.0) in developed Pacific ex-Japan. We see the return spread between the two sub-regions is widening, especially since the credit crisis, where the emerging Asia model has generated an impressive return of 14.0% pa since 2008 but the developed Pacific ex-Japan model had a flattish performance of 2.9% pa during the same period.

Rank by value spread Z-score

Rank by style 6-month return momentum

8 StylesSize

MomentumYield

ValuationRevisionGrowth

ProfitabilityRisk

Long-term relative-value of style

E/P & B/P spread rank score

1st Style

Short-term relative-strength of style

Style momentum rank score

2nd Style

3rd Style

4th Style

5th Style

6th Style

7th Style

8th Style

Combined style ranking

if previous month's market return is +ve,

60%/40% on style momentum/value spread

if previous month's market return is -ve,

40%/60% on style momentum/value spread

Our objective is to realistically implement the style-picking strategy for optimal performance in emerging Asia

Our style selection model has delivered better performance in emerging Asia than in the developed Pacific ex-Japan; the return spread between the two sub-regions is widening since the credit crisis


21

Fig. 28: Performance of style selection models for developed Pacific ex Japan and emerging Asia

Note: The return of the model is calculated as the average long-short return of the two styles selected by the model, rebalancing monthly. The style turnover is defined as the average percentage of one-way style switch each month.


In the chart below, we examine the long-side alpha and short-side alpha of the style selection model in emerging Asia. The results show that the model is more effective on the long-side, with a long-side excess return of 6.9% pa and an IR of 2.0, against a short-side excess return of 5.3% pa and an IR of 1.7. This is crucial to us as individual stock liquidities may be low and short-selling facilities are limited in some emerging markets. The effective long-side alpha provides us the flexibility to implement our style selection model in emerging Asia, not only in a long-short approach, but also as a long-only strategy with consistent outperformance over the benchmark.

Fig. 29: Emerging Asia style selection model (long-side alpha versus short-side alpha)

Note: Long-side alpha is the average of the long-only return of the two styles selected by the model, minus benchmark return. Short-side alpha is the benchmark return, minus the average of the short-only return of the two styles. We use MSCI EM Asia Index as the benchmark.


0

20

40

60

80

100

120

140

160

No

v-0

0

No

v-0

1

No

v-0

2

No

v-0

3

No

v-0

4

No

v-0

5

No

v-0

6

No

v-0

7

No

v-0

8

No

v-0

9

No

v-1

0

No

v-1

1

(%)


Emerging Asia

Developed Pacific ex‐JP Emerging Asia

Annualised return (%) 6.63 12.23

Standard deviation (%) 6.46 5.84

Information ratio 1.03 2.09

Style turnover (%) 39.6 34.4

-10

0

10

20

30

40

50

60

70

80

90

No

v-0

0

No

v-0

1

No

v-0

2

No

v-0

3

No

v-0

4

No

v-0

5

No

v-0

6

No

v-0

7

No

v-0

8

No

v-0

9

No

v-1

0

No

v-1

1

(%)

Long-side alpha

Short-side alpha

Long‐side alpha Short‐side alpha

Annualised return (%) 6.92 5.31

Standard deviation (%) 3.43 3.12

Information ratio 2.02 1.70

Results show a better long-side alpha in emerging Asia region, allowing us to implement our model as a long-only strategy


22

Optimised long-only portfolio

In this section, we apply the emerging Asia style selection model to the portfolio management process, by taking the two chosen styles each month as a source of alpha, with pre-defined turnover and other constraints. To avoid short-selling limitations in some emerging markets, we run this process in a long stock / short benchmark approach. We attempt to construct an optimised long-only portfolio in emerging Asia, which utilises the alphas from our style selection model, with restricted turnover and risk control, and is able to be implemented practically.

As an example, we did a historical simulation since end-January 2001 using the following parameters:

Fig. 30: Summary of portfolio optimization simulation

Universe MSCI Emerging Market Asia universe

Sample period January 2001 to March 2012 on a monthly basis

Benchmark MSCI Emerging Market Asia Index (total return)

Method Long stocks – short benchmark portfolio optimisation (Axioma Robust Optimisation)

Alpha factor (normalised)

Two style scores selected by our style selection model on EM Asia

Objective Maximise alpha

Risk control (annualised)

Maximum 8% on active risk

Turnover control (1-way)

Maximum 20%

Portfolio size Maximum 100 stocks on the long side

Country and sector neutral

Yes

Asset weight control Maximum 10% and minimum 0.1% on each holding

Asset active weight control

Maximum +/- 2%

Rebalancing End of every month

Performance Subsequent one-month total return in USD


The chart below presents the back test results of the portfolio optimisation process. Over the back-testing period, the optimised long-only portfolio delivered an excess return of 8.1% pa over the MSCI Emerging Asia Index, with an annualised active risk of 5.5%, leading to a return/risk ratio of 1.5 for the alpha before considering trading costs (or 1.0 after transaction costs)

A practical portfolio optimisation simulation with turnover and risk control plus other constraints

With an annualised active risk of 5.5%, the optimised long-only strategy has delivered an excess return of 8.1% pa and an IR of 1.5


23

Fig. 31: Cumulative performance of optimised emerging Asia style selection portfolios

Note: Portfolio return is total return (with dividend re-invested) in USD. Benchmark is MSCI EM Asia Total Return USD Index. Returns after transaction costs are estimated using assumed one-way costs of 50bps.

Source: Axioma, Nomura Quantitative Strategies

We also examine the country and sector effects in this simulation. By just setting either country-neutral or sector-neutral in optimisation but relaxing turnover and risk constraints, we study the performance impacts on this strategy in emerging Asia. As shown below, we note that the annualised excess return from the portfolio is higher when we relax restrictions on country and sector active weights. The same happens on the portfolio active risk. Notably, the country-neutral portfolio outperforms the sector-neutral portfolio in absolute terms, whereas it achieves a lower risk over the benchmark, such that a country-neutral approach has generated a more consistent return than sector-neutral approach, with the return/risk ratio of 1.45 versus 1.29. We believe the alpha return from our style selection model has been offset by the country and sector factors, and this result is in line with the finding in our attribution analysis, where country effect still dominates industry effect in emerging Asia.

0

10

20

30

40

50

60

70

80

90

100

Jan

-01

Jan

-02

Jan

-03

Jan

-04

Jan

-05

Jan

-06

Jan

-07

Jan

-08

Jan

-09

Jan

-10

Jan

-11

Jan

-12

(%)

Optimised long portfolio excess return

Excess return after transaction costs

Optimised long

portfolioMSCI EM Asia Excess return

Excess return after

transaction costs

Annualised return (%) 21.69 13.74 8.05 5.59

Standard derviation (%) 26.41 24.72 5.54 5.49

Information ratio 0.82 0.56 1.46 1.02

Average turnover (1‐way; %) 20 ‐ 20 20

Average holdings 98 ‐ 98 98

Country-neutral approach has generated a more consistent return than sector-neutral


24

Fig. 32: Optimised portfolio excess returns: country / sector effect

Note: We relaxed turnover and risk control during this optimization. The return is excess return of the portfolio over benchmark with different settings of country/sector neutral constraints and excludes transaction costs.


Turnover rate matters

By lowering the turnover rate, one may achieve a lower transaction costs but also reduce the power of alpha at the same time. The chart below reveals an analysis of the impact on the after-costs return with different controls on the turnover rate. Our simulation suggests the optimal one-way turnover control is 20% annualised.

0

20

40

60

80

100

120

140

160

180

200

Jan

-01

Jan

-02

Jan

-03

Jan

-04

Jan

-05

Jan

-06

Jan

-07

Jan

-08

Jan

-09

Jan

-10

Jan

-11

Jan

-12

(%)

Country and sector neutral

No control on country/sector

Country neutral

Sector neutral

Country and

sector neutral

No control on

country/sectorCountry neutral Sector neutral

Annualised return (%) 10.11 14.45 12.55 11.09

Standard derviation (%) 7.00 10.30 8.64 8.58

Information ratio 1.44 1.40 1.45 1.29

Average turnover (1‐way; %) 53 56 55 55

Average holdings 91 79 85 86

Our simulation suggests the optimal one-way turnover control is 20% annualised


25

Fig. 33: Cumulative performance with different turnover controls

Note: Actual turnover may be higher than the limit when it violates other constraints. Excess returns after transaction costs are estimated using assumed one-way costs of 50bps.


This example may serve as one of the practical ways to implement our style selection model in emerging Asia. Investors can design a systematic investment approach according to their specific risk budget, turnover control and other investment conditions, applying the alpha scores generated from our style selection model into their portfolio management process, to seek additional alpha.

-10

0

10

20

30

40

50

60

70

80

Jan

-01

Jan

-02

Jan

-03

Jan

-04

Jan

-05

Jan

-06

Jan

-07

Jan

-08

Jan

-09

Jan

-10

Jan

-11

Jan

-12

(%)5%

10%

20%

30%

40%

No limits

Turnover constraint 5% 10% 20% 30% 40% No limits

Annualised return (%) 2.28 2.81 5.59 5.45 5.10 3.52

Standard derviation (%) 3.72 4.57 5.49 6.15 6.26 6.78

Information ratio 0.61 0.62 1.02 0.89 0.81 0.52

Average turnover (1‐way; %) 9 12 20 30 38 53

The style selection model and alpha scores can be applied to portfolio management process to practically implement our model in emerging Asia


26

Appendix I: Methodology of return attribution analysis Each beta coefficient is estimated according to the following procedure. Analyses are conducted every year using weekly returns (about 52 weeks of data).

1. Multifactor model for attribution analysis As mentioned in the second section of this report, we assume that returns of individual stocks are explained by the multifactor model below.

itktIijt

Cit

Miiit ICMR

Here, Rit stands for the return of stock i denominated in local currency at time t. Mt refers to the return of the FTSE All-World Index, while Cjt and Ikt are the country and global industry factor returns at time t, respectively.

2. Pre-estimation of coefficient against global market,βMi We first estimate βMi using the regression method below.

ittMiiit uMR

3. Estimation of factor returns for country and sector factors Using βMi derived in above, we calculate the residual return, rit.

ititMiitit uMRr

Then, we estimate the weekly country factor, Cjt , and the industry factor, Ikt , by cross-sectional regression techniques.

itk

ktIki

jjt

Cjiit IdCdr

9

1

13

1 4. Estimation of coefficient (factor beta) Finally, we consider the coefficient (factor beta) for each stock by applying the country and industry factors calculated using the formulae in 3 above.

itktIijt

Cit

Miiit ICMR


27

Earnings surprise forecasting strategy • The 12/3 full-year results announcement season will soon be upon us.

• We attempt to predict which stocks will produce an earnings surprise, mainly on the basis of the continuity of individual stocks' surprise bias.

1. Earnings surprise forecasting strategy

Japanese companies’ 12/3 full-year results announcement season is approaching. Companies' profit forecasts might already be on the conservative side, as a result of downward revisions since the beginning of this year, while the economic environment has taken a turn for the better recently. Because of this, we think actual profits might come in above company forecasts—in other words, positive surprises are likely. Share price performance tends to be relatively strong for a long time after a positive surprise.

Analysis of historical data reveals that there has been a positive bias to the difference between actual profits and company forecasts in the past—ie, actual profits have been higher than company forecasts for many companies. This might be partly a reflection of the conclusion set out in Suda and Hanaeda (2008), which carries out an analysis on the basis of listed companies' responses to questions about their financial reports, and comes to the conclusion that companies see the profit forecasts they have set themselves as targets that they have to reach, and try to reach these targets even when the operating environment is difficult, for example by cutting advertising expenditure and capex or selling assets.

On the basis of this analysis, we divided companies into groups in terms of the size of the difference between their actual profits and company forecasts, and looked at the share price response of each group after their respective results announcements. We found that, when actual profits are slightly higher than the company forecast, there is often no share price response (no surprise), but that when actual profits are lower than the company forecast, even only slightly, there tends to be a sharp fall in share price. It might be that market participants have factored in the positive bias and take a harsh view when a company has not been able to meet its targets, assuming that there must be some other reason for this. It thus appears that investors should target companies that are likely to overshoot their profit forecasts substantially, and avoid companies that might undershoot their forecasts, even by only a very small margin.

In this report we introduce a method of predicting which stocks are likely to produce a positive surprise, using a quant model. Specifically, we calculate the probability of an earnings surprise using an ordered probit model in which historical forecast bias, progress through Q3 (the percentage of the full-year profit forecast achieved in first three quarters), and the absence/presence of forecast revisions are the explanatory variables. Use of this model might increase the probability of choosing stocks that will see a positive earnings surprise and thus make it possible to achieve high returns by investing in these companies before they announce their results.

1.1 Distribution of full-year earnings surprises

(1) Results often come in higher than company forecasts First of all, we define earnings surprise on the basis of the difference between actual recurring profits in the preceding fiscal year and the company's recurring profit forecast just before the results announcement, according to the formula below. Henceforth, unless otherwise specified, profits means recurring profits and forecast means company forecast.

Surprise (Surpt) =

(actual recurring profits in fiscal year t) – (company's recurring profit forecast for fiscal year t immediately before results announcement)

company’s recurring profit forecast for fiscal year t immediately before results announcement

Akihiro Murakami

+81 3 6703 1746

[email protected]

Naoko Kato

+81 3 6703 3912

[email protected]

Previously published on 19 March 2012


28

The universe is TSE-1 companies with March fiscal year-ends and the sample period is 2000 through 2011. We divided the universe into four groups on the basis of the size of the earnings surprise—+5% or more, 0–+5% (0% or more but less than +5%), -5–0% (-5% or more but less than 0%), and less than -5%, and calculated the number of stocks in each group as a percentage of the total number of stocks in the universe.

Figure 34 shows the results of our analysis. Because companies can revise their profit forecasts during the fiscal year, actual profits are often more or less in line with company forecasts. However, it is clear from Figure 34 that the distribution of earnings surprises—ie, the respective percentages of companies overshooting and undershooting their own forecasts—is not symmetrical. Actual recurring profits were greater than company forecasts for more than 60% of companies in all fiscal years in the sample period except 08/3. Even if we look only at companies for which profits were more or less as expected—ie, the -5–0% (slightly lower than expected) and 0–5% (slightly higher than expected) groups, we can see that there are more than twice as many stocks in the latter as in the former group.

Fig. 34: Distribution of earnings surprises: comparison of actual recurring profits and company forecasts (1)

Note: Universe is TSE-1 companies with March fiscal year-ends. Shows percentage of companies in universe in each group, by size of surprise, for each fiscal year. Averages are simple averages for 00/3 through 11/3.

Source: Nomura, based on data provided by each company

(2) Market participants know that earnings surprises have a positive bias Next, we looked at how market participants view this positive bias in earnings surprises. We carried out an event study to look at differences in the share price response of companies with different sizes of earnings surprise.

The universe was TSE-1 companies with March fiscal year-ends that released full-year results from 2000 through 2011. We divided the universe up into groups on the basis of the size of the earnings surprise—+5% or more, 0–+5%, -5–0%, and less than -5%—and looked at the share price response of each group. Defining the time immediately after the close of trading on the day the company announced its results as the event date (Day 0), we calculated the average excess return versus the TOPIX at each point from event date -15 to event date +120.

Figure 35 shows the results of our analysis. Stocks with a surprise factor of +5% or more show consistently strong performance, while the share price performance of companies with a surprise factor of 0–+5%—ie, where recurring profits came in slightly above company forecasts—is more or less neutral. We think this is probably because market participants regard these companies' results as "no surprise". Meanwhile, in the case of companies where recurring profits came in lower than company forecasts, share prices underperformed not only for those for which recurring profits came in 5% or more below the company forecast, but also for those where recurring profits came in only slightly below target (-5–0%).

0

10

20

30

40

50

60

70

80

90

100

00/3 01/3 02/3 03/3 04/3 05/3 06/3 07/3 08/3 09/3 10/3 11/3 Average

(%)

(yy/m)

Less than -

+5% or mor

0– +5%

-5–0%

Distribution of earnings surprises (comparison of actual recurring profits and company forecasts)


29

Share prices thus appear to factor in the positive bias in earnings surprises. It might also be the case that, when a company fails to achieve its target, market participants think this must be due to some problem, and therefore take a negative view. This seems to indicate that investors should target stocks that are likely to overshoot their profit forecasts substantially, and avoid stocks that are likely to undershoot their targets even by a small margin.

Fig. 35: Performance by size of full-year earnings surprise

Note: Universe is TSE-1 companies that announced full-year results from 2000 through 2011. Shows results of event study where close of trading on day of full-year results announcement is event day (Day 0). We divided the universe into four groups on the basis of the size of the surprise factor (+5% or more, 0–+5%, -5–0%, -5% or less) and calculated cumulative excess return versus the TOPIX (return on each group minus return on whole universe) from event day -15 through event day +120, rebased so that cumulative excess return on event day is 0%.

Source: Nomura

1.2 Forecasting which stocks are likely to see positive and negative earnings surprises

(1) Investigation based on rank correlation Next, we considered whether companies at which recurring profits come in higher or lower than company forecasts have distinguishing features, looking in particular at:

• Continuity of direction of surprise bias over time; and

• Correlation between progress through end of previous quarter and full-year earnings surprise.

First, we looked at the continuity of the direction of the surprise bias. If the positive bias in earnings surprises is due partly to the extent of companies’ own efforts, such as cost cutting for example, and also partly to discretionary decisions by management, companies that in the past have opted for a positive bias might do the same in future. Conversely, companies that are not interested in outperforming their own forecasts might not see a positive bias in future. Moreover, Tamura and Shimizu (2006) points out that there might be a bias in a company's initial forecasts as a result of a decision on the part of its management, and that this kind of bias tends to be ongoing. In other words, companies that produce optimistic (or conservative) forecasts tend to continue to produce optimistic (or conservative) forecasts. This indicates that any bias is likely to be maintained.


30

In this report, therefore, we calculate the average size of the surprise bias over the past four fiscal years, defined according to the following formula.

We then look at the correlation between earnings progress (cumulative recurring profits in Q1 through Q3 as a percentage of the company's full-year forecast) and full-year earnings surprises. While there is a tendency for management bias in profit forecasts, there is also a relationship between full-year results and profits through Q3. When the rate of progress relative to the profit forecast just before the full-year results announcement is high, if the profit forecast is not subsequently revised, the actual profit is likely to be higher than the forecast. We define progress as follows.

(2) Investigation using ordered probit model Next, we used an ordered probit model to distinguish statistically between positive earnings surprises and negative earnings surprises, on the basis of historical data. With historical surprise bias, progress through the most recent quarter, and flags indicating the presence of upward or downward revisions during the fiscal year in question (1 if there was a revision, 0 if there was not, in both cases) as explanatory variables, and response variables of 1 for the positive surprise group, 2 for the neutral group, and 3 for the negative surprise group, we used an ordered probit model to investigate the explanatory power of the explanatory variables, as shown in Figure 36.

We also used progress as an explanatory variable from 2005 onwards, as sufficient quarterly data are available from this point on. In addition, on the basis of the results set out in 1.1 (3), we define the positive surprise group as stocks for which actual profits were +5% or more higher than the company forecast, the neutral group as stocks for which actual profits were 0–+5% higher than the company forecast, and the negative surprise group as stocks for which actual profits were 0% or more lower than the company forecast.

Fig. 36: Ordered probit model

Source: Nomura

Surp < -5%

-5% <= Surp < 0%

0%<= Surp < +5%

+5% <= Surp

Negative (3)

Positive (1)

Neutral (2)

Progress (Prog) =

(actual recurring profit in Q1–Q3 in fiscal year t) – (company forecast of recurring profit in fiscal year t just before results announcement)

(company forecast of recurring profit in fiscal year t just before results announcement)

_14 ,

4

1


31

Figure 37 shows the results generated by our model. The left side shows the results of our analysis using historical surprise bias and absence/presence of downward or upward revisions during the fiscal year in question as explanatory variables, while the right side shows the results of our analysis when progress through the most recent quarter is added as an explanatory variable.

The regression coefficients for historical surprise bias are statistically significant positives, indicating that stocks with a greater positive surprise bias in the past are more likely to see positive surprises in the future. The situation varies from year to year, but stocks with upward revisions during the fiscal year also tended to see a positive surprise. In addition, while downward revisions do not have as much explanatory power as upward revisions, they do tend to mean a negative surprise. When progress is also added as an explanatory variable for 2005 onwards, the regression coefficients are positive, although the degree of statistical significance varies from year to year in this case too.

Fig. 37: Ordered probit model results

Note: Universe is TSE-1 companies that announced full-year results from 2002 through 2011. Response variables are "1" for surprise factor of +5% or more, "2" for surprise factor of 0–+5%, and "3" for surprise factor of less than 0%. In Model 1, the explanatory variables are the average earnings surprise over the past four fiscal years (SURP_bias) and flags indicating the absence/presence of upward revisions and downward revisions during the fiscal year in question, while model 2 also includes progress through the most recent quarter (PROG) as an explanatory variable. t-values are for the null hypothesis where the regression coefficient is zero. *** indicates significance at 10%, ** indicates significance at 5%, * indicates significance at 1%.

Source: Nomura

1.3 Surprise forecasts and full-year earnings surprise strategy

Using the model investigated in 1.2 above, we will now attempt to come up with a strategy that is an improvement on a regular earnings surprise strategy, as specified below.

The universe is TSE-1 stocks with March fiscal year-ends and the sample period is May 2000 through February 2012. We use the rebalancing method shown in Figure 38 (1). At the beginning of April each year, before the start of the full-year results announcement season for companies with March fiscal year-ends, we predict which companies are likely to produce positive and negative surprises. We go long on an equally weighted portfolio comprising stocks with a probability of 40% or more of producing a positive surprise, and short on an equally weighted portfolio comprising stocks with a probability of 40% or more of producing a negative surprise. We confirm the accuracy of the model's forecasts at the end of May, by which time almost all companies have

Model 1 Model 2

yy/m SURP_bias REV_up REV_dwn SURP_bias PROG REV_up REV_dwn

02/4 0.15 0.05 0.06

4.74 **** 0.33 0.64

03/4 0.12 0.36 -0.20

3.84 **** 2.60 *** -1.97 **

04/4 0.19 0.33 -0.13

5.90 **** 3.34 **** -1.39

05/4 0.11 0.25 -0.17 0.16 0.18 0.27 -0.05

3.77 **** 2.60 *** -1.87 * 3.24 *** 0.98 2.13 *** -0.38

06/4 0.09 0.27 -0.15 0.11 0.47 0.27 -0.25

2.94 *** 3.06 *** -1.65 * 3.16 *** 3.58 **** 2.74 *** -2.25 ***

07/4 0.15 0.13 -0.27 0.14 0.56 0.12 -0.22

5.04 **** 1.37 -2.90 *** 4.14 **** 4.11 **** 1.21 -2.03 ***

08/4 0.17 0.29 -0.19 0.19 0.36 0.23 -0.11

5.58 **** 2.56 *** -2.25 *** 5.34 **** 3.06 *** 1.91 * -1.18

09/4 0.12 0.33 -0.10 0.09 0.08 0.32 -0.03

4.10 **** 3.27 *** -1.52 2.26 *** 1.43 2.90 *** -0.39

10/4 0.10 0.28 -0.24 0.13 0.03 0.39 -0.29

3.49 *** 3.48 **** -2.55 *** 3.18 *** 0.85 4.21 **** -2.45 ***

11/4 0.24 0.39 -0.25 0.25 0.25 0.27 -0.03

7.49 **** 4.01 **** -2.08 *** 6.97 **** 2.54 *** 2.63 *** -0.20


32

announced their results. We stay long in stocks that produced a positive surprise, as expected, but remove stocks that failed to produce a positive surprise as expected, and replace them with stocks that produced a positive surprise, although the model had not predicted this. We then hold these stocks through March in the following year. We do the same for stocks that produce a negative surprise. For the purpose of comparison, we also measure the performance of a regular full-year earnings surprise strategy. This regular full-year earnings surprise strategy involves going long on stocks that produced a positive surprise, as is clear by end-May, when almost all companies have announced their results, and short on stocks that produced a negative surprise, and holding these positions for one year (Figure 38 (2)).

Fig. 38: Analysis methodology

Source: Nomura

Figure 39 shows the results of our analysis. The top graph shows statistical data for cumulative return and the bottom graph for monthly return (average return, standard deviation (risk), and average/standard deviation are annualized).

The surprise forecasting model + full-year earnings surprise strategy (1) produced an average return of 4.85%, which is higher than the average return of 3.61% on the regular full-year earnings surprise strategy (2). In particular, (1) produced an average return of 14.48% in April, the month before the results announcement season, with a standard deviation of 7.20% and an average/standard deviation of 2.01. (2), meanwhile, produced an average return in April of -2.31%. In our view, this indicates that our surprise forecasting model is effective.

Forecast accurate hold stockInvest before end-May on basisof surprise forecasts Forecast accurate replace stock

Invest and hold untilnext FY results on basis of surprise forecasts

Invest for follow ing year immediately afterresults announcements

(1) Surprise forecasting model + full-year earnings surprise strategy

(2) Full-year earnings surprise strategy

Mar Jun Jul DecAugApr MayResults

announcements

FebJan Mar Jun Jul AugApr MayResults

announcements

• • •

Mar Jun Jul DecAugApr MayResults

announcements

FebJan Mar Jun Jul AugApr MayResults

announcements

• • •


33

Fig. 39: Results of grouping simulation

Note: Using our surprise forecasting model, we divided up the universe of TSE-1 companies on the basis of the probability of producing a positive or negative surprise as of the beginning of April each year. We then measured the performance of a strategy of going long on the stocks with a positive surprise probability of 40% or more and going short on the stocks with a negative surprise probability of 40% or more, and holding these positions through end-May, then dividing the universe into three groups on the basis of the size of the surprise factor, as calculated from their most recent results, at the beginning of June, and going long on an equally weighted portfolio of stocks with the highest value of the surprise factor and short on an equally weighted portfolio of stocks with the lowest value of the surprise factor, and holding these positions for the next 10 months. t-values are for the null hypothesis where the regression coefficient is zero. *** indicates significance at 10%, ** indicates significance at 5%, * indicates significance at 1%.

Source: Nomura

Reference materials:

• Akihiro Murakami,2012,” Japanese equity quantitative strategy: earnings surprise forecasting strategy”, Nomura Securities Global Quantitative Research report (15 March 2012)

• Hiromichi Tamura, Yasuhiro Shimizu, 2006, “Company forecast errors and mispricing,” Nomura Securities Global Quantitative Research report (26 July 2006)

• Kazuyuki Suda, Hideki Hanaeda, 2008, “Nihon kigyo no zaimu hokoku,” Securities Analysts Journal (May 2008)

• John R. Graham, Campbell R. Harvey, and Shiva Rajgopal, “Value Destruction and Financial Reporting Decisions,” Financial Analysts Journal, November/December 2006

0

10

20

30

40

50

60

70

00/5 01/5 02/5 03/5 04/5 05/5 06/5 07/5 08/5 09/5 10/5 11/5

(end-00/5 = 0%)

(yy/m)

Cumulative return

Surprise forecasting model + full-year earnings surprise strategy (1)

Full-year earnings surprise strategy (2)

Statistical data for monthly return (average and standard deviation data are annualized)

Strategy Average Standard deviation

t-value Average/

standard deviation

(1) Surprise forecasting model + full-year earnings surprise strategy

Entire period 4.85 3.48 4.79 *** 1.39 of which, April 14.48 7.20 2.01

(2) Full-year earnings surprise strategy Entire period 3.67 2.86 4.39 *** 1.28 of which, April -2.31 2.22 -1.04


34


35


36


37

Appendix A-1

Analyst Certification

We, Inigo Fraser-Jenkins, Sandy Lee and Akihiro Murakami, hereby certify (1) that the views expressed in this Research report accurately reflect our personal views about any or all of the subject securities or issuers referred to in this Research report, (2) no part of our compensation was, is or will be directly or indirectly related to the specific recommendations or views expressed in this Research report and (3) no part of our compensation is tied to any specific investment banking transactions performed by Nomura Securities International, Inc., Nomura International plc or any other Nomura Group company.


38

Important Disclosures Online availability of research and conflict-of-interest disclosures Nomura research is available on www.nomuranow.com, Bloomberg, Capital IQ, Factset, MarkitHub, Reuters and ThomsonOne. Important disclosures may be read at http://go.nomuranow.com/research/globalresearchportal/pages/disclosures/disclosures.aspx or requested from Nomura Securities International, Inc., on 1-877-865-5752. If you have any difficulties with the website, please email [email protected] for help. The analysts responsible for preparing this report have received compensation based upon various factors including the firm's total revenues, a portion of which is generated by Investment Banking activities. Unless otherwise noted, the non-US analysts listed at the front of this report are not registered/qualified as research analysts under FINRA/NYSE rules, may not be associated persons of NSI, and may not be subject to FINRA Rule 2711 and NYSE Rule 472 restrictions on communications with covered companies, public appearances, and trading securities held by a research analyst account. Any authors named in this report are research analysts unless otherwise indicated. Industry Specialists identified in some Nomura International plc research reports are employees within the Firm who are responsible for the sales and trading effort in the sector for which they have coverage. Industry Specialists do not contribute in any manner to the content of research reports in which their names appear. Marketing Analysts identified in some Nomura research reports are research analysts employed by Nomura International plc who are primarily responsible for marketing Nomura’s Equity Research product in the sector for which they have coverage. Marketing Analysts may also contribute to research reports in which their names appear and publish research on their sector. Distribution of ratings (US) The distribution of all ratings published by Nomura US Equity Research is as follows: 39% have been assigned a Buy rating which, for purposes of mandatory disclosures, are classified as a Buy rating; 0% of companies with this rating are investment banking clients of the Nomura Group*. 54% have been assigned a Neutral rating which, for purposes of mandatory disclosures, is classified as a Hold rating; 0% of companies with this rating are investment banking clients of the Nomura Group*. 7% have been assigned a Reduce rating which, for purposes of mandatory disclosures, are classified as a Sell rating; 0% of companies with this rating are investment banking clients of the Nomura Group*. As at 31 March 2012. *The Nomura Group as defined in the Disclaimer section at the end of this report. Distribution of ratings (Global) The distribution of all ratings published by Nomura Global Equity Research is as follows: 46% have been assigned a Buy rating which, for purposes of mandatory disclosures, are classified as a Buy rating; 40% of companies with this rating are investment banking clients of the Nomura Group*. 43% have been assigned a Neutral rating which, for purposes of mandatory disclosures, is classified as a Hold rating; 45% of companies with this rating are investment banking clients of the Nomura Group*. 11% have been assigned a Reduce rating which, for purposes of mandatory disclosures, are classified as a Sell rating; 21% of companies with this rating are investment banking clients of the Nomura Group*. As at 31 March 2012. *The Nomura Group as defined in the Disclaimer section at the end of this report. Explanation of Nomura's equity research rating system in Europe, Middle East and Africa, US and Latin America The rating system is a relative system indicating expected performance against a specific benchmark identified for each individual stock. Analysts may also indicate absolute upside to target price defined as (fair value - current price)/current price, subject to limited management discretion. In most cases, the fair value will equal the analyst's assessment of the current intrinsic fair value of the stock using an appropriate valuation methodology such as discounted cash flow or multiple analysis, etc. STOCKS A rating of 'Buy', indicates that the analyst expects the stock to outperform the Benchmark over the next 12 months. A rating of 'Neutral', indicates that the analyst expects the stock to perform in line with the Benchmark over the next 12 months. A rating of 'Reduce', indicates that the analyst expects the stock to underperform the Benchmark over the next 12 months. A rating of 'Suspended', indicates that the rating, target price and estimates have been suspended temporarily to comply with applicable regulations and/or firm policies in certain circumstances including, but not limited to, when Nomura is acting in an advisory capacity in a merger or strategic transaction involving the company. Benchmarks are as follows: United States/Europe: Please see valuation methodologies for explanations of relevant benchmarks for stocks (accessible through the left hand side of the Nomura Disclosure web page: http://go.nomuranow.com/research/globalresearchportal);Global Emerging Markets (ex-Asia): MSCI Emerging Markets ex-Asia, unless otherwise stated in the valuation methodology. SECTORS A 'Bullish' stance, indicates that the analyst expects the sector to outperform the Benchmark during the next 12 months. A 'Neutral' stance, indicates that the analyst expects the sector to perform in line with the Benchmark during the next 12 months. A 'Bearish' stance, indicates that the analyst expects the sector to underperform the Benchmark during the next 12 months. Benchmarks are as follows: United States: S&P 500; Europe: Dow Jones STOXX 600; Global Emerging Markets (ex-Asia): MSCI Emerging Markets ex-Asia. Explanation of Nomura's equity research rating system in Japan and Asia ex-Japan STOCKS Stock recommendations are based on absolute valuation upside (downside), which is defined as (Target Price - Current Price) / Current Price, subject to limited management discretion. In most cases, the Target Price will equal the analyst's 12-month intrinsic valuation of the stock, based on an appropriate valuation methodology such as discounted cash flow, multiple analysis, etc. A 'Buy' recommendation indicates that potential upside is 15% or more. A 'Neutral' recommendation indicates that potential upside is less than 15% or downside is less than 5%. A 'Reduce' recommendation indicates that potential downside is 5% or more. A rating of 'Suspended' indicates that the rating and target price have been suspended temporarily to comply with applicable regulations and/or firm policies in certain circumstances including when Nomura is acting in an advisory capacity in a merger or strategic transaction involving the subject company.

http://www.nomuranow.com/�

http://go.nomuranow.com/research/globalresearchportal/pages/disclosures/disclosures.aspx�

http://go.nomuranow.com/research/globalresearchportal�


39

Securities and/or companies that are labelled as 'Not rated' or shown as 'No rating' are not in regular research coverage of the Nomura entity identified in the top banner. Investors should not expect continuing or additional information from Nomura relating to such securities and/or companies. SECTORS A 'Bullish' rating means most stocks in the sector have (or the weighted average recommendation of the stocks under coverage is) a positive absolute recommendation. A 'Neutral' rating means most stocks in the sector have (or the weighted average recommendation of the stocks under coverage is) a neutral absolute recommendation. A 'Bearish' rating means most stocks in the sector have (or the weighted average recommendation of the stocks under coverage is) a negative absolute recommendation. Target Price A Target Price, if discussed, reflect in part the analyst's estimates for the company's earnings. The achievement of any target price may be impeded by general market and macroeconomic trends, and by other risks related to the company or the market, and may not occur if the company's earnings differ from estimates.


40

Disclaimers This document contains material that has been prepared by the Nomura entity identified at the top or bottom of page 1 herein, if any, and/or, with the sole or joint contributions of one or more Nomura entities whose employees and their respective affiliations are specified on page 1 herein or identified elsewhere in the document. Affiliates and subsidiaries of Nomura Holdings, Inc. (collectively, the 'Nomura Group'), include: Nomura Securities Co., Ltd. ('NSC') Tokyo, Japan; Nomura International plc ('NIplc'), UK; Nomura Securities International, Inc. ('NSI'), New York, US; Nomura International (Hong Kong) Ltd. (‘NIHK’), Hong Kong; Nomura Financial Investment (Korea) Co., Ltd. (‘NFIK’), Korea (Information on Nomura analysts registered with the Korea Financial Investment Association ('KOFIA') can be found on the KOFIA Intranet at http://dis.kofia.or.kr ); Nomura Singapore Ltd. (‘NSL’), Singapore (Registration number 197201440E, regulated by the Monetary Authority of Singapore); Capital Nomura Securities Public Company Limited (‘CNS’), Thailand; Nomura Australia Ltd. (‘NAL’), Australia (ABN 48 003 032 513), regulated by the Australian Securities and Investment Commission ('ASIC') and holder of an Australian financial services licence number 246412; P.T. Nomura Indonesia (‘PTNI’), Indonesia; Nomura Securities Malaysia Sdn. Bhd. (‘NSM’), Malaysia; Nomura International (Hong Kong) Ltd., Taipei Branch (‘NITB’), Taiwan; Nomura Financial Advisory and Securities (India) Private Limited (‘NFASL’), Mumbai, India (Registered Address: Ceejay House, Level 11, Plot F, Shivsagar Estate, Dr. Annie Besant Road, Worli, Mumbai- 400 018, India; Tel: +91 22 4037 4037, Fax: +91 22 4037 4111; SEBI Registration No: BSE INB011299030, NSE INB231299034, INF231299034, INE 231299034, MCX: INE261299034); NIplc, Dubai Branch (‘NIplc, Dubai’); NIplc, Madrid Branch (‘NIplc, Madrid’) and NIplc, Italian Branch (‘NIplc, Italy’). THIS MATERIAL IS: (I) FOR YOUR PRIVATE INFORMATION, AND WE ARE NOT SOLICITING ANY ACTION BASED UPON IT; (II) NOT TO BE CONSTRUED AS AN OFFER TO SELL OR A SOLICITATION OF AN OFFER TO BUY ANY SECURITY IN ANY JURISDICTION WHERE SUCH OFFER OR SOLICITATION WOULD BE ILLEGAL; AND (III) BASED UPON INFORMATION FROM SOURCES THAT WE CONSIDER RELIABLE, BUT HAS NOT BEEN INDEPENDENTLY VERIFIED BY NOMURA GROUP. Nomura Group does not warrant or represent that the document is accurate, complete, reliable, fit for any particular purpose or merchantable and does not accept liability for any act (or decision not to act) resulting from use of this document and related data. To the maximum extent permissible all warranties and other assurances by Nomura group are hereby excluded and Nomura Group shall have no liability for the use, misuse, or distribution of this information. Opinions or estimates expressed are current opinions as of the original publication date appearing on this material and the information, including the opinions and estimates contained herein, are subject to change without notice. Nomura Group is under no duty to update this document. Any comments or statements made herein are those of the author(s) and may differ from views held by other parties within Nomura Group. Clients should consider whether any advice or recommendation in this report is suitable for their particular circumstances and, if appropriate, seek professional advice, including tax advice. Nomura Group does not provide tax advice. Nomura Group, and/or its officers, directors and employees, may, to the extent permitted by applicable law and/or regulation, deal as principal, agent, or otherwise, or have long or short positions in, or buy or sell, the securities, commodities or instruments, or options or other derivative instruments based thereon, of issuers or securities mentioned herein. Nomura Group companies may also act as market maker or liquidity provider (as defined within Financial Services Authority (‘FSA’) rules in the UK) in the financial instruments of the issuer. Where the activity of market maker is carried out in accordance with the definition given to it by specific laws and regulations of the US or other jurisdictions, this will be separately disclosed within the specific issuer disclosures. This document may contain information obtained from third parties, including ratings from credit ratings agencies such as Standard & Poor’s. Reproduction and distribution of third party content in any form is prohibited except with the prior written permission of the related third party. Third party content providers do not guarantee the accuracy, completeness, timeliness or availability of any information, including ratings, and are not responsible for any errors or omissions (negligent or otherwise), regardless of the cause, or for the results obtained from the use of such content. Third party content providers give no express or implied warranties, including, but not limited to, any warranties of merchantability or fitness for a particular purpose or use. Third party content providers shall not be liable for any direct, indirect, incidental, exemplary, compensatory, punitive, special or consequential damages, costs, expenses, legal fees, or losses (including lost income or profits and opportunity costs) in connection with any use of their content, including ratings. Credit ratings are statements of opinions and are not statements of fact or recommendations to purchase hold or sell securities. They do not address the suitability of securities or the suitability of securities for investment purposes, and should not be relied on as investment advice. Any MSCI sourced information in this document is the exclusive property of MSCI Inc. (‘MSCI’). Without prior written permission of MSCI, this information and any other MSCI intellectual property may not be reproduced, re-disseminated or used to create any financial products, including any indices. This information is provided on an "as is" basis. The user assumes the entire risk of any use made of this information. MSCI, its affiliates and any third party involved in, or related to, computing or compiling the information hereby expressly disclaim all warranties of originality, accuracy, completeness, merchantability or fitness for a particular purpose with respect to any of this information. Without limiting any of the foregoing, in no event shall MSCI, any of its affiliates or any third party involved in, or related to, computing or compiling the information have any liability for any damages of any kind. MSCI and the MSCI indexes are services marks of MSCI and its affiliates. Investors should consider this document as only a single factor in making their investment decision and, as such, the report should not be viewed as identifying or suggesting all risks, direct or indirect, that may be associated with any investment decision. Nomura Group produces a number of different types of research product including, among others, fundamental analysis, quantitative analysis and short term trading ideas; recommendations contained in one type of research product may differ from recommendations contained in other types of research product, whether as a result of differing time horizons, methodologies or otherwise. Nomura Group publishes research product in a number of different ways including the posting of product on Nomura Group portals and/or distribution directly to clients. Different groups of clients may receive different products and services from the research department depending on their individual requirements. Figures presented herein may refer to past performance or simulations based on past performance which are not reliable indicators of future performance. Where the information contains an indication of future performance, such forecasts may not be a reliable indicator of future performance. Moreover, simulations are based on models and simplifying assumptions which may oversimplify and not reflect the future distribution of returns. Certain securities are subject to fluctuations in exchange rates that could have an adverse effect on the value or price of, or income derived from, the investment. The securities described herein may not have been registered under the US Securities Act of 1933 (the ‘1933 Act’), and, in such case, may not be offered or sold in the US or to US persons unless they have been registered under the 1933 Act, or except in compliance with an exemption from the registration requirements of the 1933 Act. Unless governing law permits otherwise, any transaction should be executed via a Nomura entity in your home jurisdiction. This document has been approved for distribution in the UK and European Economic Area as investment research by NIplc, which is authorized and regulated by the FSA and is a member of the London Stock Exchange. It does not constitute a personal recommendation, as defined by the FSA, or take into account the particular investment objectives, financial situations, or needs of individual investors. It is intended only for investors who are 'eligible counterparties' or 'professional clients' as defined by the FSA, and may not, therefore, be redistributed to retail clients as defined by the FSA. This document has been approved by NIHK, which is regulated by the Hong Kong Securities and Futures Commission, for distribution in Hong Kong by NIHK. This document has been approved for distribution in Australia by NAL, which is authorized and regulated in Australia by the ASIC. This document has also been approved for distribution in Malaysia by NSM. In Singapore, this document has been distributed by NSL. NSL accepts legal responsibility for the content of this document, where it concerns securities, futures and foreign exchange, issued by their foreign affiliates in respect of recipients who are not accredited, expert or institutional investors as defined by the Securities and Futures Act (Chapter 289). Recipients of this document in Singapore should contact NSL in respect of matters arising from, or in connection with, this document. Unless prohibited by the provisions of Regulation S of the 1933 Act, this material is distributed in the US, by NSI, a US-registered broker-dealer, which accepts responsibility for its contents in accordance with the provisions of Rule 15a-6, under the US Securities Exchange Act of 1934. This document has not been approved for distribution in the Kingdom of Saudi Arabia (‘Saudi Arabia’) or to clients other than 'professional clients' in the United Arab Emirates (‘UAE’) by Nomura Saudi Arabia, NIplc or any other member of Nomura Group, as the case may be. Neither this document nor any copy thereof may be taken or transmitted or distributed, directly or indirectly, by any person other than those authorised to do so into Saudi Arabia or in the UAE or to any person located in Saudi Arabia or to clients other than 'professional clients' in the UAE. By accepting to receive this document, you represent that you are not located in Saudi Arabia or that you are a 'professional client' in the UAE and agree to comply with these restrictions. Any failure to comply with these restrictions may constitute a violation of the laws of Saudi Arabia or the UAE. NO PART OF THIS MATERIAL MAY BE (I) COPIED, PHOTOCOPIED, OR DUPLICATED IN ANY FORM, BY ANY MEANS; OR (II) REDISTRIBUTED WITHOUT THE PRIOR WRITTEN CONSENT OF A MEMBER OF NOMURA GROUP. If this document has been distributed by electronic transmission, such as e-mail, then such transmission cannot be guaranteed to be secure or error-free as information could be intercepted, corrupted, lost, destroyed, arrive late or incomplete, or contain viruses. The sender therefore does not accept liability for any errors or omissions in the contents of this document, which may arise as a result of electronic transmission. If verification is required, please request a hard-copy version. Nomura Group manages conflicts with respect to the production of research through its compliance policies and procedures (including, but not limited to, Conflicts of Interest, Chinese Wall and Confidentiality policies) as well as through the maintenance of Chinese walls and employee training. Additional information is available upon request and disclosure information is available at the Nomura Disclosure web page: http://go.nomuranow.com/research/globalresearchportal/pages/disclosures/disclosures.aspx

http://go.nomuranow.com/research/globalresearchportal/pages/disclosures/disclosures.aspx�

Documents

Nomura Global Quantitative Research Monthly 04-2012 510050