Asset Allocation with Exchange-Traded Funds: From Passive to Active Management. Felix Goltz

Transcription

1 Asset Allocation with Exchange-Traded Funds: From Passive to Active Management Felix Goltz

2 1. Introduction and Key Concepts 2. Using ETFs in the Core Portfolio so as to design a Customized Allocation Consistent with Institutional Investors Preferences and Constraints 3. Using ETFs in the Satellite Portfolios so as to Generate Outperformance Based on Active Asset Allocation Decisions 4. Putting the Pieces Together

3 Asset management is the only field where professionals do worse than amateurs The average fund manager under-performs the index There is no evidence of persistence for those who over-perform Possible explanations (other than lack of skill) 1. Managers focus too much on stock picking and not enough on asset allocation 2. Tracking error constraints: Not enough distinction between alpha (active) and beta (passive) management

4 Factors Explaining Dynamics of Returns over Long Horizons 2.1% 4.6% 1.8% 91,5% Strategic Asset Allocation Market Timing Others Stock Picking - Asset allocation explains a large fraction of the variability of returns of an average investor Drivers of Differences in Performance between Mutual Funds % 3.5% 40. Strategic Asset Allocation Tactical Asset Allocation Stock Picking Fees - Asset allocation decisions explain most of the difference in returns between actively managed funds

5 Traditional active management does not favour a clear distinction between Beta management Alpha management From the CAPM, a manager s performance can be decomposed into ( ) = + ( ) [ ] α β + Exposure to additional systematic risk factors should be taken into account

6 Separating alpha from beta management In case of a purely passive investment strategy ( ), the performance is only based on beta management In case of a market (factor)-neutral hedge fund (beta=0), performance beyond risk-free rate only emanates from active bets Most active managers are mostly passive (tracking error constraints) Such a confusing mélange des genres is likely to be costly and inefficient A paradigm change: The core-satellite approach =

7 Example: Cost reduction for an International Equity portfolio ( 100m, 4% tracking error) Fees with traditional approach: 100bp = 1m Fees with core-satellite approach: 36bp = 360,000 «Core» Core «Satellite» Satellite Global Weight TrackingError 2 4% (x.80+2x.20) Management Fees 20 bps 100 bps 36 bps (20x x.20)

8 2. ETFs in the Core Portfolio

9 The dimensions of diversification Mixing stocks and bonds Mixing different segments of the equity markets (international, style and sector), and different segments of the bond markets (short vs. long maturities, investment grade vs speculative grades) Objective: Find the best possible trade-off between risk and return Risk may be measured in terms of Volatility (measure of average risk) VaR (measure of extreme risk)

10 8 0 F r e q u e n c y R e t u r n s % % % - 6 % - 2 % 2 % 6 % 1 0 % 1 4 % 1 8 % M o r e VaR is a forecast of a given percentile (in the lower tail) of the distribution of returns on a portfolio over some period.

11 ! Normal vs non normal return distribution 0.35% 0.3 Probability Normal distribution 0.25% Asset return distribution % % Returns

12 " Kurtosis: frequency of large positive or negative asset returns Skewness: frequency with which large returns in a particular direction occur Kurtotic and normal distributions Skewed and normal distributions kurtotic normal skewed normal return return

13 # Historical VaR Simple parametric Gaussian approach: based on normality in the portfolio return distribution = µ σ =" α! where µ = mean return, σ = volatility, α = confidence level (99%, 97% or 95%) Semi-parametric approach: a good trade-off Example of Cornish-Fisher VaR: = + " #! + #& " $ $! $ "# $ %! # Where S = skewness et K = excess kurtosis = µ σ

14 We look at the following two cases: (A) Stock investor : Portfolio allocation: 10 Stocks (B) Balanced investor : Portfolio allocation: 4 Stocks, 6 Bonds We perform a minimisation of VaR at 95% (rolling window analysis with rebalancing every six months based on daily data) Rolling window analysis with rebalancing every six months Daily returns data First calibration period: 01/1999 to 12/2001 First out of sample period 01/2002 to 06/2002, last out of sample period: 07/2004 to 12/2004

15 $ A. Stock Investor 1. Base Case: 10 in Euro Stoxx Diversification by size: Allocation between Euro Stoxx 50, Euro Stoxx Midcap, Euro Stoxx Small Cap [weight constraints: min 1, max 5] 3. Diversification by country: Allocation between Euro Stoxx 50, S&P 500, FTSE 100, MSCI Japan [weight constraints: min 3, max 5 for Euro Stoxx, min 1, max 5 for the rest]. 4. Diversification by size and country: Allocation between Euro Stoxx 50, Euro Stoxx Midcap, Eurostoxx Small Cap, S&P 500, FTSE 100, MSCI Japan) [weight constraints as in 3 with: min 3, max 5 for the total allocation to Euro Stoxx indices] B. Balanced Investor, 1. Base case: 4 Stocks (DJ Euro Stoxx 50), 6 Bonds (IBOXX Euro Sovereign) Approaches 2, 3 and 4 as above while the bond holdings are held constant

16 %#& ' ( DJ EURO STOXX 50 DJ EURO STOXX MID DJ EURO STOXX SMALL FTSE 100 S&P 500 MSCI JAPAN IBOXX EURO SOVERE IGN Stock Investor Base Case 10 Size Country % 33% Size and Country % 1 25% 28% Balanced Investor Base Case 4 6 Size 4% 16% 2 6 Country 12% 4% 11% 13% 6 Size and Country 4% 4% 9% 4% 9% 1 6 Mean weights for the portfolios formed over the period 01/ /2004. Weights remain very stable (sometimes unchanged) over all different approaches

17 &( Mean Return Volatility VaR (95%) Volatility Improvement VaR Improvement Stock Investor Base Case 0.42% 26.32% 44.2 Size 1.77% 16.28% 31.67% 38% 28% Country 0.76% 16.18% 30.39% 39% 31% Size and Country 1.07% 14.23% 27.62% 46% 38% Balanced Investor Base Case % 10.08% 19.88% Size 0.36% 6.22% 13.36% 38% 33% Country % 6.26% 12.99% 38% 35% Size and Country 0.16% % 45% 41% Summary Statistics for the in-sample period 01/ /2004

18 &( Mean Return Volatility VaR (95%) Volatility Improvement VaR Improvement Stock Investor Base Case -4.36% 27.38% 44.88% Size 5.05% 15.58% 29.38% 43% 35% Country 1.15% 16.36% 30.28% 4 33% Size and Country 3.41% 14.08% 26.78% 49% 4 Balanced Investor Base Case -0.58% 10.06% 19.61% Size 3.22% 5.55% 11.65% 45% 41% Country 1.68% 5.87% 12.22% 42% 38% Size and Country 2.66% 5.03% 10.57% 5 46% Summary Statistics for the out-of-sample period 01/ /2004

19 2 VaR (95%) 15% 1 Size and Country Country Size Base Case 4% 6% 8% 1 Volatility Risk measures for different approaches during the out-of-sample period 01/ /2004. For the case of the balanced investor

20 )* Active asset allocation allows for a decrease in portfolio risk without a corresponding decrease in returns thanks to the benefits of diversification Return Core Portfolio (Here) ' ' ' Broad commercial index Not necessarily an efficient portfolio Optimal strategic asset allocation decisions implemented with ETFs allow investors to achieve optimal location on the efficient frontier Objective: generate lowest possible risk for given expected return Risk Average active manager (There) Underperforms commercial index (taking into account costs and fees)

21 3. ETFs in the Satellite Portfolio

22 * + Asset returns are to some extent predictable Campbell (2000): "Most financial economists appear to have accepted that aggregate returns do contain an important predictable component. Timing strategies between styles, sectors or countries may be implemented Such a Global Tactical Asset Allocation (GTAA) program has been shown to dominate stock picking in terms of performance Specialised tactical managers in the US include Wells Capital Management, First Quadrant, PanAgora Example: Timing strategy between equity styles (see Amenc, Malaise, Martellini and Sfeir (2003))

23 ,$* Equity styles have contrasted performance under different economic conditions /97 05/97 09/97 01/98 05/98 09/98 01/99 05/99 09/99 01/00 05/00 09/00 01/01 05/01 09/01 01/02 05/ /97 05/97 09/97 01/98 05/98 09/98 01/99 05/99 09/99 01/00 05/00 09/00 01/01 05/01 09/01 01/02 05/02 S&P 500 GROWTH - S&P 500 VALUE S&P 600 SMALL CAP - S&P 500 Style Differentials (Annualised Returns)

24 %+! Some financial variables have a natural impact on stock returns For each style differential, we generate a list of variables based on an economic analysis and then use econometric analysis to construct predictive models Predictive variables can be found within the following broad categories Interest rates Risk Valuation Ratios Stock returns Others (liquidity, commodity prices, FX rates)

25 *$%# Monthly Allocation of the Style Timing Strategy Recommendations Large Cap Growth Large Cap Value Large Cap Small Cap June % 24.86% 21.72% % July % 21.42% August % % 28.75% September % 38.35% 11.17% % October % 21.31% % November % 41.85% % 14.41% December % 28.78% % 29.2 January % 30.02% % 30.12% February % % 29.53% March % 27.87% 23.18% % April % % May % 30.53% % 29.38% June % 30.71% % 29.65% July % 30.63% % 29.57% August % September % 44.66% October % 15.16% % 44.62% November % December % 30.52% % 29.22% January % 15.08% % 43.27% February % March % 42.69% 12.33% % April May % June % % 10.73% % July % % % 14.44% August % % % September % % % October % 26.27% 22.96% % November % % 48.56% December % 38.42% 10.69% %

26 *$ Summary Statistics: Characteristics of the TSA Strategy January February March April May June July August September October November December % 1.93% 3.97% % 0.22% 4.32% % 1.84% % 0.74% 0.95% -1.54% % -0.25% 1.99% % % 0.41% % % -0.56% 1.87% 1.99% 0.3 TSA Fund S&P 500 Cumulative Return => Excess Return Annualised Return % Annualised Std Deviation 4.71% 18.72% => TSA Volatility 4 X lower than S&P500's Downside Deviation (3.) 2.26% 11.49% Sortino (3.) Sharpe st Centile -1.55% % % Negative Returns 22.58% 61.29% => Few losing months with TSA Strategy Worst Monthly Drawdown -1.56% Peak to Valley -1.56% % => No outliers with TSA Strategy Months in Max Drawdown 1 25 Months to recover 1 no recovery Beta => TSA Strategy uncorrelated with stocks indices Alpha => Significant Alpha

27 4. Putting the Pieces Together: Using ETFs in a Static or Dynamic Core-Satellite Approach

28 # Above Core: Strategic Asset Allocation with ETFs Satellite: Tactical Asset Allocation with ETFs Below Allocation between the core and the satellite may vary over time This allows for asymmetrically managing the Tracking Error of the global portfolio See Amenc, Malaise and Martellini (2004)

29 # Tracking error is not necessarily bad - Good tracking error: out-performance with respect to the benchmark - Bad tracking error: underperformance with respect to the benchmark Dynamic Allocation between the core and satellite - Full access to good tracking error - Maintaining the level of bad tracking error below a given threshold

30 * This is an extension of CPPI to a relative return context The method leads to an increase in the weight of the satellite when the satellite has outperformed the core Accumulation of past out-performance results in an increase in the cushion, and in the potential for a more aggressive (higher tracking error) strategy in the future Underperformance of the satellite leads to a decrease of the weight of the satellite portfolio (tighter tracking error constraint).

31 , -*, 8 0 F r e q u e n c y & % % -1 0 % -6 % - 2 % 2 % 6 % 1 0 % 1 4 % 1 8 % M o re!!!"# $ #% *% 8 0 F re q u e n c y %# & % -1 4 % -1 0 % -6 % -2 % 2 % 6 % 1 0 % 1 4 % 1 8 % M o re!!!"# $ ( (( )

32 1 %2/3 )1 % )# # $ & % ( % #( #% : ) #* #;%; #% $(% < &% %( %#% %& <, $&( $&& $% $ ( <= (#* (&# (%* (*$ <= ":>6! (& (# (#; ($% <+/= ($% (% (* (% <+/=":>6! (& ( * (( # 8 (** (*% (*# (*(? < #@$@#(($ #* %%$##;%;&&(;#%% *%$(% $&%& 1 #@$@#(($ %&#( %&#( %&#( %&#( A ;(((## ;$(; (#%(% ((#&& +, -, " - ). ),!- ) / - + ) - - ),- 0,

33 % Satellite Weighting Cumulative Excess Return (bps) / 12/ 21/ 30/ 08/ 17/ 25/ 09/ 18/ 26/ 05/ 16/ 30/ 10/ 18/ 27/ 11/ 23/ 03/ 11/ 20/ 31/ 11/ 25/ 03/ 12/ 0.012/ 03/ 05/ 07/ 10/ 12/ 02/ 05/ 07/ 09/ 12/ 02/ 04/ 07/ 09/ 11/ 02/ 04/ 07/ 09/ 11/ 01/ 04/ 06/ 09/ 11/ Satellite Weighting Excess Returns < "#% &%!- /"!

34 +" * ETFs allow implementing asset allocation decisions since they offer exposure to systematic risk factors (no security selection) ETFs are often regarded as vehicles for implementing passive strategies The benefits of ETFs are actually much larger: Strategic Asset Allocation Tactical Asset Allocation Dynamic core-satellite allocation approach

35 )*. Return Core + Satellite Active investment decisions (alphas): aim at shifting the efficient frontier Objective: outperform the reference benchmark for a given level of risk ' ' ' ' ' Core Portfolio Optimal strategic asset allocation decisions (betas): allow investors to achieve optimal location on the efficient frontier Objective: generate lowest possible risk for given expected return Active satellite Portable alpha approach: only alphas, no betas Commercial index Not necessarily an efficient portfolio Risk Average active manager Underperforms commercial index (taking into account costs and fees)