DATA61 Dynamic Asset Allocation Strategy

Transcription

1 DATA61 Dynamic Asset Allocation Strategy Z. Zhu, R. Zhang, N. Langrené, G. Lee CSIRO e Publish: EP162465

2 Copyright Commonwealth Scientific and Industrial Research Organisation To the extent permitted by law, all rights are reserved and no part of this publication covered by copyright may be reproduced or copied in any form or by any means except with the written permission of CSIRO. Important disclaimer CSIRO advises that the information contained in this publication comprises general statements based on scientific research. The reader is advised and needs to be aware that such information may be incomplete or unable to be used in any specific situation. No reliance or actions must therefore be made on that information without seeking prior expert professional, scientific and technical advice. To the extent permitted by law, CSIRO (including its employees and consultants) excludes all liability to any person for any consequences, including but not limited to all losses, damages, costs, expenses and any other compensation, arising directly or indirectly from using this publication (in part or in whole) and any information or material contained in it. CSIRO is committed to providing web accessible content wherever possible. If you are having difficulties with accessing this document please contact

3 Introduction Conventional asset allocation strategies commonly used in the asset management industry are normally categorised as: 1). defensive, 2). conservative, 3) balanced, 4) growth, and 5) high growth. These passive portfolio strategies are formulated to cater to the varying degrees of risk preferences of investors. Such passive strategies can bear significant downside risk when the market goes bearish, for example, the conservative strategy was not a safe haven when the global market dropped during To achieve consistent higher returns in a challenging environment, we have developed a methodology for dynamically allocating asset classes through a Real options approach. In this paper, we present the dynamic asset allocation strategy which can be shown to substantially enhance portfolio returns, particularly in a volatile investment environment, in contrast to the passive six asset allocation strategies often used in the industry. Dynamic asset allocation As an example, we use the dynamic asset allocation approach for a portfolio investing in five asset classes: Australian shares, international shares, emerging market shares, Australian bond and gold. We will show that such a dynamic strategy can outperform the six standard passive asset allocation strategies commonly used in the industry. We consider a monthly rebalancing portfolio invested in the following five markets: ASSET CLASS ETF PROVIDER Domestic Australian Shares Vanguard VAS ETF International Shares ishares IOO ETF Emerging Market Shares ishares IEM ETF Bonds ishares IAF ETF Gold ETF Securities GOLD ETF For the dynamic asset allocation strategy, the portfolio weight on each asset is chosen optimally on a monthly interval over the 12 months. We assume that an investor s objective is to maximize the expected wealth at the end of the 12 months. The transaction cost when rebalancing a portfolio is assumed to be 0.3% proportional to the portfolio turnover value. The initial wealth is assumed to be $1,000,000. Two dynamic allocation strategies are deployed: 1. UNBOUNDED DYNAMIC: the portfolio weight can assume any value from 0% to 100% 2. BOUNDED DYNAMIC: the portfolio weight is restricted within the range 10% to 50% The two dynamic strategies are benchmarked against the six standard portfolio strategies commonly adopted in the asset management industry for different risk tolerances: ASSET CLASS PRESERVATION DEFENSIVE CONSERVATIVE BALANCED GROWTH HIGH GROWTH VAS.AX 32% 34% 35% 36% 37% 37% IOO.AX 18% 19% 20% 21% 22% 25% IEM.AX 0% 5% 10% 15% 18% 20% IAF.AX 30% 25% 20% 15% 13% 10% GOLD.AX 20% 17% 15% 13% 10% 8% Dynamic Asset Allocation Strategy 1

4 Comparison of performances We use two indicators to demonstrate performances by the dynamic asset allocation strategy: 1. Back testing of market data: real monthly market data from January 2015 to January Expected performance: out of sample simulation results based on forecasted 10,000 possible future scenarios from January 2015 to January Back testing of market data The table below shows the returns for the real market data input from January 2015 to January 2016 for each of the eight strategies: STRATEGY RETURN ON REAL MARKET DATA UNBOUNDED DYNAMIC 14.95% BOUNDED DYNAMIC 0.16% PRESERVATION 2.15% DEFENSIVE 3.31% CONSERVATIVE 4.20% BALANCED 5.19% GROWTH 5.78% HIGH GROWTH 6.15% The graph below shows the evolution of the portfolio wealth value (normalized) for each strategy according to the real market data from January 2015 to January BOUNDED DYNAMIC UNBOUNDED DYNAMIC PRESERVATION DEFENSIVE CONSERVATIVE BALANCED GROWTH HIGH GROWTH 2 Dynamic Asset Allocation Strategy

5 As shown clearly in the graph, the unbounded dynamic strategy outperforms the six standard asset allocation strategies of the industry, which all generated negative returns, with the High Growth strategy producing a negative 6.15%. The unbounded dynamic (UNBOUNDED DYNAMIC) strategy can put 100% portfolio weight on any single asset classes, but produced a total return of 14.95%. In contrast, the bounded dynamic strategy (BOUNDED DYNAMIC) generated a return of 0.16%, still a much better performance than the most conservative strategy: the Preservation portfolio which produced a return of 2.15% during the period. The table below provides detailed monthly ETF prices, optimal monthly portfolio weights (called optimal decisions in Real Options methodology) of the BOUNDED DYNAMIC strategy and the corresponding monthly portfolio values generated for the real market data from January 2015 to January NORMALISED ETF PRICES BOUNDED DYNAMIC STRATEGY Portfolio Weights WEALTH VAS.AX IOO.AX IEM.AX IAF.AX GOLD.AX VAS.AX IOO.AX IEM.AX IAF.AX GOLD.AX 2/01/ % 10% 50% 20% 10% /02/ % 50% 10% 20% 10% /03/ % 10% 50% 20% 10% /04/ % 20% 10% 10% 10% /05/ % 10% 10% 20% 10% /06/ % 10% 10% 10% 20% /07/ % 20% 10% 10% 50% /08/ % 10% 10% 10% 20% /09/ % 10% 50% 10% 20% /10/ % 50% 10% 20% 10% /11/ % 50% 20% 10% 10% /12/ % 10% 32% 10% 38% /01/ The table below shows the same monthly ETF prices, optimal monthly portfolio weights of the UNBOUNDED DYNAMIC strategy and the corresponding monthly portfolio values generated for the real monthly market data input from January 2015 to January NORMALIZED ETF PRICES UNBOUNDED DYNAMIC STRATEGY Portfolio Weights WEALTH VAS.AX IOO.AX IEM.AX IAF.AX GOLD.AX VAS.AX IOO.AX IEM.AX IAF.AX GOLD.AX 2/01/ % 0% 100% 0% 0% /02/ % 100% 0% 0% 0% /03/ % 0% 100% 0% 0% /04/ % 0% 0% 0% 0% /05/ % 0% 0% 0% 0% /06/ % 0% 0% 0% 0% /07/ % 0% 0% 0% 100% /08/ % 0% 0% 0% 0% /09/ % 0% 100% 0% 0% /10/ % 100% 0% 0% 0% /11/ % 100% 0% 0% 0% /12/ % 0% 0% 0% 100% /01/ Expected performance Dynamic Asset Allocation Strategy 3

6 To calculate the expected performances for the period of January/2015 January/2016, we simulated 10,000 future possible scenarios for the five underlying asset classes. For each forecasted future scenario, the six standard industry investment strategies and the bounded/unbounded dynamic strategies each produced a corresponding final portfolio value at January It should be pointed out, the portfolio values at January 2016 for the six passive strategies were simply calculated by summing up the weighted returns of the five underlying assets. However, for the dynamic strategies, their portfolio values at January 2016 could not be computed in such a simple manner, because the optimal monthly portfolio weights were different for different scenarios of the underlying assets. For example, the optimal monthly portfolio weights listed in the previous two tables for the bounded/unbounded dynamic strategies were computed for the real monthly market data input, and for different forecast scenarios of the five underlying assets, the dynamic strategies would generate different monthly portfolio weights. For each of the eight different investment strategies, its 10,000 possible final portfolio values at January 2016 were used to calculate key performance indicators for this strategy, as summarised in the table below: STRATEGY EXPECTED RETURN VOLATILITY 5% VaR PROB[RETURN<0] UNBOUNDED DYNAMIC 53.92% 17.93% 27.51% 0.01% BOUNDED DYNAMIC 31.22% 9.75% 16.23% 0.01% PRESERVATION 11.65% 5.62% 2.87% 1.28% DEFENSIVE 13.08% 5.99% 3.73% 0.93% CONSERVATIVE 14.42% 6.46% 4.32% 0.85% BALANCED 15.76% 7.00% 4.79% 0.80% GROWTH 16.71% 7.25% 5.39% 0.71% HIGH GROWTH 17.79% 7.61% 5.94% 0.64% It should be noted that the high expected volatility of the UNBOUNDED DYNAMIC strategy constituted mainly from the potential for large upside gains. The downside risks of both dynamic strategies were in fact very small when compared with the six standard portfolio strategies commonly used in the industry (as seen from the probability for negative returns). The graph below displays the density distributions for the returns of the two dynamic strategies and the Growth strategy adopted commonly in the industry: 4 Dynamic Asset Allocation Strategy

7 Conclusion The dynamic asset allocation strategy presented here can outperform the commonly adopted portfolio strategies in the industry, such as the superannuation sector. For the real market data between January 2015 and January 2016, the unbounded dynamic strategy (the portfolio weights can be 100% on a single asset) generated a 14.95% return, whereas the High Growth strategy used in the industry generated a 6.15% return. The bounded dynamic strategy (the portfolio weights are restricted between 10% and 50%) generated 0.16%, which was still a much better return than all the six standard strategies commonly used in the industry. For the 10,000 out of sample simulations, a much higher expected returns were achieved by the dynamic strategies than by the six standard portfolio strategies used in the industry, additionally, these higher returns were achieved with reduced downside risk of making a loss (as seen from the reduced probability of making negative returns). In this paper, we have only presented the dynamic asset allocation strategy of targeting for maximum portfolio returns at the end of 12 months. Different targeted returns and risk profiles can be prescribed for the dynamic asset allocation strategy, and the dynamic strategy will generate different optimal monthly portfolio weights to achieve the prescribed target portfolio performances. This dynamic allocation strategy can also be applied to creating optimal currency hedging ratios when foreign asset holdings need to be hedged either to maximise the total return or to reduce volatility of such a portfolio of foreign assets. Dynamic Asset Allocation Strategy 5

8 CONTACT US t e csiroenquiries@csiro.au w AT CSIRO, WE DO THE EXTRAORDINARY EVERY DAY We innovate for tomorrow and help improve today for our customers, all Australians and the world. Our innovations contribute billions of dollars to the Australian economy every year. As the largest patent holder in the nation, our vast wealth of intellectual property has led to more than 150 spin off companies. With more than 5,000 experts and a burning desire to get things done, we are Australia s catalyst for innovation. CSIRO. WE IMAGINE. WE COLLABORATE. WE INNOVATE. FOR FURTHER INFORMATION DATA61 Dr Zili Zhu Team Leader for Real Options and Financial Risk Decision Sciences Program t +61 (03) e Zili.Zhu@csiro.au fordecisions/financial services 6 Dynamic Asset Allocation Strategy