Axioma Insight Quarterly Risk Review Third-Quarter 2015 Chinese Edition Analysis Date September 30, 2015 Melissa Brown, CFA mbrown@axioma.com Diana Rudean, PhD drudean@axioma.com Natan Borshansky nborshansky@axioma.com 2015 Axioma, Inc. Axioma Insight Quarterly Risk Review 1
What s Inside Focus on the Chinese Market Third-Quarter 2015 Overview 5 Factor Returns and Volatility 9 Implications for Tracking Error 13 Correlations and Cross-Sectional Dispersion 14 Sector and Stock Specific Risk 16 Figure 1. Short-Horizon Fundamental Model Predicted Volatility 4 Figure 2. Medium-Horizon Fundamental Model Predicted Volatility 4 Figure 3. Predicted Risk, Four Model Variants 5 Figure 4. Components of Risk, China Fundamental Models 5 Figure 5. Total Risk Spread, CSI 300 6 Figure 6. Short-Medium Spread, CSI 300 6 Figure 7. Statistical-Fundamental Spread, CSI 300 7 Figure 8. Change-In-Risk Decomposition, CSI 300 8 Figure 9. 12-Month Cumulative Factor Returns, China Fundamental Model 9 Figure 10. Minimum Active Risk Varied by Number of Names 13 Figure 11. Median Realized Asset-Asset Correlations and Percent Negative 14 Figure 12. Cross-Sectional Asset and Factor Dispension 14 Figure 13. Sector Predicted Risk and Contribution to Overall Risk 16 Table 1. Benchmark Predicted Volatility 3 Table 2. Monthly Style Factor Returns 9 Table 3. Style Factor Predicted Risk 10 Table 4. Industry Predicted Risk 11 Table 5. Style-Factor Correlations 15 Table 6. Biggest Changes in Medium-Horizon Fundamental Model Total Risk 17 2015 Axioma, Inc. Axioma Insight Quarterly Risk Review 2
Table 1. Benchmark Predicted Volatility, Axioma Fundamental and Macroeconomic¹ Models 2 All benchmarks, at both horizons, saw risk increase, many substantially, during Q3 * FTSE Europe minus FTSE Eurobloc **FTSE Developed Europe minus FTSE Eurobloc ***Portugal, Italy, Ireland, Greece and Spain from FTSE Eurobloc Source: FTSE, Russell Investments, Standard and Poor s, China Securities Index Company Ltd., Axioma Inc. ¹ US only ²Some of the markets Axioma covers do not have official benchmarks, but we are able to back into the risk estimates. For example, FTSE does not have a Euro Crisis index, but we are able to create a derivative by starting with the FTSE Developed Europe index and removing all countries except Portugal, Italy, Ireland, Greece and Spain. The remaining stocks are then reweighted in proportion to their original weights. In this report we will refer to these derivations as benchmarks, although we will only use the provider s name (e.g., FTSE) for their official indices. Source: FTSE, Russell Investments, Standard and Poor s, China Securities Index Company Ltd., Axioma Inc. 2015 Axioma, Inc. Axioma Insight Quarterly Risk Review 3
Figure 1. Short-Horizon Fundamental Model Predicted Volatility China s risk similar to most other benchmarks a year ago stood out by the end of Q3 Figure 2. Medium-Horizon Fundamental Model Predicted Volatility In an unusual twist, Russell 2000 risk fell below Russell 1000 at the end of Q3 * FTSE Europe minus FTSE Eurobloc **FTSE Developed Europe minus FTSE Eurobloc ***Portugal, Italy, Ireland, Greece and Spain from FTSE Eurobloc Source: FTSE, Russell Investments, Standard and Poor s, China Securities Index Company Ltd., Axioma Inc. 2015 Axioma, Inc. Axioma Insight Quarterly Risk Review 4
Focus on the Chinese Market Third-Quarter 2015 Overview Figure 3. Predicted Risk, Four Model Variants, CSI 300 China s risk soared in Q2 and continued to rise in Q3 Figure 4. Components of Risk, China Fundamental Models, CSI 300 Style risk has been extremely variable 2015 Axioma, Inc. Axioma Insight Quarterly Risk Review 5
Figure 5. Total Risk Spread, CSI 300 Model disagreement far exceeds that of any of the markets we cover closely The total risk spread is the difference between the highest and lowest risk forecasts from the four model variants (statistical, fundamental, short-horizon and medium-horizon). Figure 6. Short-Medium Spread, CSI 300 as short-horizon forecasts remained high above medium-horizon 2015 Axioma, Inc. Axioma Insight Quarterly Risk Review 6
Figure 7. Statistical-Fundamental Spread, CSI 300 and statistical forecasts were well above fundamental, suggesting risk not captured in fundamental factors 2015 Axioma, Inc. Axioma Insight Quarterly Risk Review 7
To understand what is driving the change in risk for each benchmark, we look at the contribution to change from each of the underlying components: portfolio composition/weights, portfolio stock characteristics, stock-specific volatility, factor volatility and factor correlations. The overall impact of assetlevel correlations is not seen until we combine these components, so to see the impact of changes in stock correlations we must also decompose the full asset covariance matrix. Figure 8. Change-In-Risk* Decomposition, CSI 300³ Volatilities rose substantially, but higher asset correlations also played a role in the higher risk forecast *China Medium-Horizon Fundamental Model 3 The components of a risk forecast are the portfolio holdings, its factor exposures, the covariance matrix, and stock-specific risk. In order to decompose the change in risk of a benchmark or portfolio, we employ the following methodology: We first look at the impact of the change in holdings, so we use last period s risk model with the current portfolio to calculate a risk forecast, and the difference is attributable solely to the change in holdings. Second, to evaluate the change in stocks characteristics, we update the factor exposures, and again use last period s risk model, but current holdings and exposures. Third, we look at the impact of specific risk changes by using current specific risk estimates but the prior period s covariance matrix. We calculate the impact of changes in correlation by using last period s correlations with all the other data as of the current period. Finally, the residual is the impact of the change in factor volatility. While the ordering described above will affect the results, we find that the results do not change substantially when we change the order. Intuitively we believe that the biggest impact will come from change in factor volatility, and our results bear out our intuition. 2015 Axioma, Inc. Axioma Insight Quarterly Risk Review 8
Factor Returns and Volatility Table 2. Monthly Style Factor Returns, China Fundamental Model Style factor returns were much higher/lower than average in July and August, but settled down by September *Axioma medium-horizon model, January 1999 present Highlighted cells are one standard deviation above or below average Source: Axioma, Inc. Figure 9. 12-Month Cumulative Factor Returns, China Fundamental Model Value had a strong showing for the last 12 months in China Source: Axioma, Inc. 2015 Axioma, Inc. Axioma Insight Quarterly Risk Review 9
Table 3. Style-Factor Predicted Risk, China Fundamental Models All style factors risk was in the top decile relative to history by the end of Q3 The Current Rank vs. History column shows how the quarter-end value compared with month-end values for the style factor since 1999. Source: Axioma, Inc. 2015 Axioma, Inc. Axioma Insight Quarterly Risk Review 10
Table 4. Industry Predicted Risk, China Fundamental Models...as was risk for every industry Notes: Industries with risk in the top quartile are highlighted in pink, those with risk in the bottom quartile are highlighted in green. The Current Rank vs. History column shows how the quarter-end value compares with month-end values for the industry since 1999. Source: Axioma, Inc. 2015 Axioma, Inc. Axioma Insight Quarterly Risk Review 11
Table 4. Con t. Notes: Industries with risk in the top quartile are highlighted in pink. Those with risk in the bottom quartile are highlighted in green. The Current Rank vs. History column shows how the quarter-end value compares with month-end values for the industry since 1999. Source: Axioma, Inc. 2015 Axioma, Inc. Axioma Insight Quarterly Risk Review 12
Implications for Tracking Error Axioma users are often interested in how minimum active risk for a portfolio with a given number of names might have changed as risk changed. To accomplish this we use Axioma s frontier backtest capability, in which our objective is to minimize active risk while varying the number of names in the portfolio. Figure 10. Minimum Active Risk* Varied By Number of Names, CSI 300 The frontier shifted up substantially by mid-year 2015, and continued to rise in Q3 *China Medium-Horizon Fundamental Model 2015 Axioma, Inc. Axioma Insight Quarterly Risk Review 13
Correlation and Cross-Sectional Dispersion Figure 11. Median Realized Asset-Asset Correlations and Percent of Correlation That Were Negative, CSI 300 Median correlations neared 15-year highs during Q3 Figure 12. Cross-Sectional Asset and Factor Dispersion, CSI 300 Factor and asset dispersion ended the quarter at just about their long-term average levels 2015 Axioma, Inc. Axioma Insight Quarterly Risk Review 14
Table 5. Style-Factor Correlations Changes in many factor correlations were much larger than average during Q3 Source: Axioma, Inc. 2015 Axioma, Inc. Axioma Insight Quarterly Risk Review 15
Sector and Stock-Specific Risk Figure 13. Sector Predicted Risk and Contribution to Overall Risk*, CSI 300 Risk for all sectors increased substantially in Q3, with the biggest gain in Materials *China Medium-Horizon Fundamental Model 2015 Axioma, Inc. Axioma Insight Quarterly Risk Review 16
Table 6. Biggest Changes in Medium-Horizon Fundamental Model Total Risk, CSI 300 20 Largest Increases 20 Largest Decreases Russell 1000 and Russell 2000 are registered trademarks of Russell Investments. Copyright Russell Investments 2015. All rights reserved. FTSE Japan Index, FTSE Eurobloc Index, FTSE 350 Index, FTSE Europe Index, FTSE Asia ex-japan Index, FTSE Asia Pacific Index, FTSE North America Index, FTSE Developed Index, FTSE All-World Index; and FTSE Emerging Index are trademarks of FTSE. All rights reserved. S&P, S&P INDICES, and STANDARD & POOR S are registered trademarks of Standard & Poor s Financial Services LLC. CSI 300 is a registered trademark of China Securities Index Co., Ltd. Copyright China Securities Index Co., Ltd 2015. All rights reserved. 2015 Axioma, Inc. Axioma Insight Quarterly Risk Review 17
United States and Canada: +1 212-991-4500 Europe: +44 (0)20 7856 2424 Asia: +852-8203-2790 New York Axioma, Inc. 17 State Street Suite 2700 New York, NY 10004 Phone: 212-991-4500 Fax: 212-991-4539 Hong Kong Axioma, (HK) Ltd. Unit B, 17/F, Entertainment Building 30 Queen s Road Central Hong Kong Phone: +852-8203-2790 Fax: +852-8203-2774 Singapore Axioma, (Asia) Pte Ltd. 30 Raffles Place #23-00 Chevron House Singapore 048622 Phone: +65 6233 6835 Fax: +65 6233 6891 London Axioma, (UK) Ltd. 30 Crown Place London, EC2A 4EB Phone: +44 (0) 20 7856 2451 Fax: +44 (0) 20 3006 8747 Atlanta Axioma, Inc. 400 Northridge Road Suite 850 Atlanta, GA 30350 Phone: 678-672-5400 Fax: 678-672-5401 San Francisco Axioma, Inc. 201 Mission Street Suite 2150 San Francisco, CA 94105 Phone: 415-614-4170 Fax: 415-614-4169 Geneva Axioma CH Rue du Rhone 69, Floor 2 1207 Geneva, Switzerland Phone: +33 611 96 81 53 Sales: sales@axioma.com Client Support: support@axioma.com Careers: careers@axioma.com 2015 Axioma, Inc. Axioma Insight Quarterly Risk Review 18