Japan Quantitative Strategy Post-ROE-P/B curve steeppgening strategy Nomura Securities Co., Ltd.
Akihiro Murakami Chief Quantitative Strategist, Japan See Appendix A-1 for analyst certification, important disclosures Nomura Securities Co., Ltd. and the status of non-US analysts. +81 (0)3 6703 1746 akihiro. murakami@nomura. com Any authors named on this report are research analysts unless otherwise indicated. May, 2013 Summary
Current investment environment in the Japanese equity market : ・ JittfdfdttiftJapanese equity quant fund performance and representative factors ・ Who are buying Japanese stocks?
Our investment strategy : (()a) Maj or bu yers of Ja panese e quities are nonresident investors ・ Strategy utilizing non-resident investors’ stock selection skills
(b) Reliability of analyst forecasts improving ・ Earnings revisions focusing on the ROE-P/B curve
(c) The BOJ continues to be a major buyer of ETFs ・ Contrarian to the passive fund trading
1 Who are buying Japanese stocks?
Trading by trust banks, insurance and nonresident investors
TOPIX End-2010/12 = 0% (bn Y)Yen) 40 1,750 35 1,500 30 1,250 25 Nonresidents (rhs) 1,000 20 750 15 10 500 5 250 0 0 -5 -250 -10 -500 -15 Trust banks & -750 -20 Insurance (rhs) TOPIX (lhs) -1,000 -25 -1,250 -30 Great East Japan European Resurgence of Changeof Government -35 -1,500 Earthquake Financial crisis European & -40 Financial crisis Abenomics-Driven Rally -1,750 -45 2011 Q1 Q2 Q3 Q4 2012 Q1 Q2 Q3 Q4 2013 Q1 Q2-2,000 12/6 12/9 13/3 11/3 11/6 11/9 12/3 (yyyy/m) 12/12 11/12 10/12 00 00 00 00 00 00 00 2 2 2 2 2 2 2 20 20 20
Note: Shows trust banks‘, insurance cos’ and nonresident investors' (bottom graph) trading value, taken from equity purchases by investor category data (right axis) and TOPIX return(left axis). Sample period is January 2011 through 10.May.2013 . Source: Nomura, from TSE's equity trading value by investor category data 2 As part of monetary easing, the BOJ also keeps buying Japanese ETFs
BOJ's ETF purchases and the TOPIX
TOPIX End -2010/12 = 0% (bn Yen ) 40 BOJ's ETF 40 35 purchases (rhs) 30 30 25 20 20 15 10 10 5 0 0 -5 -10 -10 -15 -20 TOPIX (lhs) -20 -25 -30 Great East Japan European Resurgence of Change of Government -30 Earthquake Financial crisis European & -35 Financial crisis Abenomics-Driven Rally -40 2011 Q1 Q2 Q3 Q4 2012 Q1 Q2 Q3 Q4 2013 Q1 Q2 -40 (yyyy/m) 010/12 011/12 012/12 2011/3 2011/6 2011/9 2012/3 2012/6 2012/9 2013/3 22 22 22
Note: Shows the actual amount of ETFs purchased by BOJ (right axis) and TOPIX return(left axis). Sample period is January 2011 through 10.May.2013 . Source: Nomura, based on BOJ and other data 3 Japanese equity quant funds started to recover since the change of government
Average performance of Japanese equity quant funds
End-2010/12 = 0% Cumulative return 10
5
0
-5
-10 Quant Fund Change of -15 Government 16th. Dec. 2012 2011 Q1 Q2 Q3 Q4 2012 Q1 Q2 Q3 Q4 2013 Q1 Q2 -20 Great East Japan European Resurgence of Earthquake Financial crisis crisis European Abenomics -DrivenRallyDriven Rally -25 Financial crisis
(yyyy/m) 2013/3 2012/3 2012/6 2012/9 2011/3 2011/6 2011/9 2012/12 2011/12 2010/12
Note: Shows daily average return (Jensen's alpha), on an indexed basis, on 23 funds (domestic publicly offered investment trusts and nonresident investor funds) that invest in Japanese equities using a quant-based approach. Sample includes long-only, individual stock long-short, and market-neutral (long on individual stocks, short on index futures, for example) funds. Sample period is January 2011 through 10.May.2013 . Source: Nomura 4 Mimic portfolio’s performance of quant funds
Average performance of Mimic portfolio and Quant funds
End-2010/12 = 0% Cumulative return 10 Mimic Quant Portfolio ( E/P + Revision + B/P + Reversal) 5
0
-5
-10 Quant Fund
-15
2011 Q1 Q2 Q3 Q4 2012 Q1 Q2 Q3 Q4 2013 Q1 Q2 -20 Great East Japan European Resurgence of Earthquake Financial crisis crisis European Abenomics-Driven Rall y -25 Financial crisis
(yyyy/m) 2013/3 2012/3 2012/6 2012/9 2011/3 2011/6 2011/9 2012/12 2011/12 2010/12
Note: Shows daily return of mimic portfolio (the composite factor of E/P, B/P, Revision and reversal(past 3month return)) and daily average return (Jensen's alpha), on an indexed basis, on 23 funds (domestic publicly offered investment trusts and nonresident investor funds) that invest in Japanese equities using a quant-based approach. Sample includes long-only, individual stock long-short, and market-neutral (long on individual stocks, short on index futures, for example) funds. Sample period is January 2011 through 10.May.2013 . Source: Nomura 5 Value factors worked as the main driver for the Q4 quant recovery But…..
Daily performance of E/P and B/P
End-2010/12 = 0% Cumulative returnreturn 20
15
10 E/P
5
0
-5
-10 B/P -15 2011 Q1 Q2 Q3 Q4 2012 Q1 Q2 Q3 Q4 2013Q1 Q2 -20 Resurgence of Great East Japan European Abenomics -DrivenRallyDriven Rally EhEart hqua kke financial crisis European -25 financial crisis 012/12 010/12 011/12 2013/3 2011/3 2011/6 2011/9 2012/3 2012/6 2012/9
22 22 22 (yyyy/ m ))
Note: Universe of TOPIX500 stocks is divided into 5 quintiles, with an equal number of stocks in each quintile, by factor value. Portfolios are rebalanced at the beginning of each month. Cumulative spread return (#5 – #1) is calculated on a daily basis (end-Dec. 2010 = 0). We did not take transaction costs into account. Analysis is based on historical share prices and does not guarantee future performance. Sample period is 4 Jan. 2011 – 10.May.2013. Source: Nomura 6 B/P factors were driven by credit risk factor
Factor performance for B/P and 5-year CDS spreads
EdEnd-2010/12 =0% 0% Cumulative return 20
10 B/P
0
-10 CDS5Y
-20
2011 Q1 Q2 Q3 Q4 2012 Q1 Q2 Q3 Q4 2013Q1 Q2 -30 Great East Japan European Resurgence of Abenomics-Driven Rally Earthquake financial crisis European financial crisis -40 010/12 011/12 012/12 2011/3 2011/6 2011/9 2012/3 2012/6 2012/9 2013/3 (yyyy/m) 22 22 22
Note: We divided the universe of TOPIX 500 stocks into five groups in terms of both B/P and 5-year CDS spreads at the beginning of each month. Figure shows cumulative monthly returns on a strategy of going long on the group of stocks with the highest factor values and short on the group of stocks with the lowest factor values, for each factor. We did not take transaction costs into account. Analysis is based on historical share prices and does not guarantee future performance. Sample period is January 2011 through 10.May.2013. Source: Nomura 7 Credit risk factor may no longer be a driver for B/P factors
CDS spreads for high-B/P and low-B/P stocks
((p)bp) CDS spreads for high‐B/P and low‐B/P stocks 400 High B/P 350
300
250
200 Low B/P 150
100 2011 Q1 Q2 Q3 Q4 2012 Q1 Q2 Q3 Q4 2013Q1 Q2 50 Great East Japan Resurgence of European Abenomics-Driven Rally Earthquake financial crisis European 0 financial crisis 010/12 011/12 012/12 2011/3 2011/6 2011/9 2012/3 2012/6 2012/9 2013/3 22 22 22
Note: Universe is TOPIX 500 stocks for which 5-year CDS spread data were available. We divided the universe into five groups on the basis of B/P factor values at the beginning of each month, and calculated the median 5-year CDS spread for the groups of stocks with the highest and lowest B/P, respectively. Sample period is January 2011 through 10.May.2013 . Source: Nomura 8 We expect profit-based factors to emerge as the next main driver
Daily performance of ROE, revision and recurring profit growth
End -2010/12 = 0% Cumulativereturn return 30
20 Revision
10 ROE
0
-10
-20 Recurring profit growth -30 2011 Q1 Q2 Q3 Q4 2012 Q1 Q2 Q3 Q4 2013Q1 Q2 Great East Japan European Resurgence of Abenomics-Driven Rally -40 Earthquake financial crisis European financial crisis crisis
-50 010/12 011/12 012/12 2011/3 2011/6 2011/9 2012/3 2012/6 2012/9 2013/3
22 22 22 (yyyy/m)
Note: Universe of TOPIX500 stocks is divided into 5 quintiles, with an equal number of stocks in each quintile, by factor value. Portfolios are rebalanced at the beginning of each month. Cumulative spread return (#5 – #1) is calculated on a daily basis (end-Dec. 2010 = 0). Sample period is 4 Jan. 2011 – 10 May.2013. We did not take transaction costs into account. Analysis is based on historical share prices and does not guarantee future performance. Source: Nomura 9 Nonresident investor purchases likely to work more positively for profit-based factors than value factors
Comparison of risk characteristics: Stocks overweighted by domestic pension funds vs. Stocks overweighted by nonresident investors
Size 0.20 Sensitivity to dollar B/P 0.10 0000.00 Historical volatility E/P (60 months) -0.10 -0.20 -0300.30 Historical beta Forecast dividend yield (60 months) -0.40
Debt ratio Sales growth
Past 1 year return ROE
Past 3-month return
Domestic pension funds #5-#1 Non-resident #5-#1 TOPIX Note: we calculated risk characteristics of each portfolio and plotted the historical average of each risk characteristic. We conducted a cross-sectional normalization of each set of data so that weighted average market cap was 0 and variance was 1 at the beginning of each month, adjusting standard deviation to ±3σ if it exceeded that level, and repeated this process three times. Sample period is June 1995 through March 2013. Source: Nomura 10 Idea 1: Learning from non-resident investors’ stock selection skills
Performance of stocks overweighted by domestic pension funds and nonresident investors
(End-1995/6 = 0%) Cumulative return 100 Active weight factor of Nonresident investors 80
60 Active weight factor of Domestic Pension 40
20
0
‐20 ROE factor
‐40 00612 00706 00712 00806 00812 00906 00912 01006 01012 01106 01112 01206 01212 99506 99512 99606 99612 99706 99712 99806 99812 99906 99912 00006 00012 00106 00112 00206 00212 00306 00312 00406 00412 00506 00512 00606 11 11 11 11 11 11 11 11 11 11 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22
Note: We divided the universe of TSE-1 stocks into five groups based on their active weightings in the portfolios of nonresident investors at the start of each month. Figure shows cumulative monthly returns on a strategy combining long positions on stocks in the highest group (overweighted stocks) with short positions on stocks in the lowest group (underweighted stocks). We did not take transaction costs into account. Analysis is based on historical share prices and does not guarantee future performance. Sample period is June 1995 through April 2013. Source: Nomura 11 Company valuation and profitability closely associated
A simple residual income model gives the relationship between future ROE and P/B ratio. ∞ , , 1 1 1 The ROE-P/B curve: the cross sectional relationshipp( between ROE and valuation (P/B)
P/B (x) Relationship between ROE and P/B 7.0 Jan 2009 Apr 2013 606.0 Fitted curve (Jan 2009)
5.0 Fitted curve (Apr 2013)
4.0
3.0
2.0
1.0
0.0 -15. 0 -10. 0 -505.0 000.0 505.0 10. 0 15. 0 20. 0 25. 0 30. 0 35. 0 ROE (%) Note: We plotted each stock's expected ROE (next-FY forecast where available, otherwise current-FY forecast) on the x-axis and its P/B on the y-axis, in a two-dimensional scatter plot. Universes were TSE-1 stocks as at the beginning of April 2013 and the beginning of January 2009, respectively. To remove outliers, we excluded stocks in the top 1% and bottom 1% of the universe in terms of ROE from our analysis. 12 Source: Nomura The ROE-P/B curve fitted by converted model
Relationship between ROE and valuation (log P/B)
The convert ed mod el : , _ , , ,
ln (P/B) Converted relationship between ROE and P/B (Apr 2013) 3.0 P/B (x) Relationship between ROE and P/B 7.0 2.0 6.0 Apr 2013 Fitted curve (Apr 2013) 5.0 101.0
4.0
0.0 3.0
y = 0.4556x + 0.078 2.0 -101.0
1.0
-2.0 0.0 -15.0 -10.0 -5.0 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 ROE (()%) -3.0 -3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 (low) Standardized ROE (high)
Note: We plotted each stock‘s standardized expected ROE on the x-axis and its log P/B on the y-axis, in a two-dimensional scatter plot. Universe was TSE-1 stocks at the beginning of April 2013. We estimated the best fit line using model , _ , where ln(*) is the natural log of *, δ and β are the intercept and regression coefficient, respectively, at time t, and ε is the , , t t i,t residual. To estimate the regression coefficient, , we standardized ROE, the explanatory variable, so that the average was 0 and the standard deviation was 1, and rounded any values beyond ±3σ to ±3σ. To remove outliers, we excluded stocks in the top 1% and bottom 1% of the universe in terms or ROE from our analysis. Source: Nomura 13 The ROE-P/B curve has recently steepened rapidly
The slope of the curve (β coefficient) has recovered to the level prior to Lehman collapse.
Time seri es β coeffic ien t for ROE-P/B curve
coefficient 0.7
0.6 The association between expected ROE and P/B is strong. 0.5 Steep
0.4
0.3
The association is weak. 0.2 Flat
0.1
0.0 198701 198801 198901 199001 199101 199201 199301 199401 199501 199601 199701 199801 199901 200001 200101 200201 200301 200401 200501 200601 200701 200801 200901 201001 201101 201201 201301 (yyyymm)