The Challenges to Market-Timing Strategies and Tactical Asset Allocation

Transcription

1 The Challenges to Market-Timing Strategies and Tactical Asset Allocation Joseph H. Davis, PhD The Vanguard Group Investment Counseling & Research and Fixed Income Groups

2 Agenda Challenges to traditional investment strategies Some expect a low return environment in years ahead Lower security-level return dispersion after the bubble Greater interest in market timing ( bet on beta to get alpha ) Macroeconomic risk strongly influences expected returns TAA requires forecasting macro trends or turning points Alpha requires a non-consensus bet on systematic risk factor Can economists and their data help? The pros and cons of market-timing strategies The evidence of timing performance is dismal, on average Explanations and evidence from real-time tactical strategies > 2

3 Market-timing and tactical asset allocation (TAA) Differences between concepts is size of active risk incurred Market-timing = all or nothing trades among asset classes Tactical asset allocation = portfolio tilts based on information Types of TAA Asset class, style, sector, global TAA Large potential rewards from market-timing strategies Bauer and Dahlquist (2001): Perfect stock-cash times yield 20%+ returns Kao and Shumaker (1999): Growth-value trades can double returns Does TAA work? Summarize the empirical evidence > 3

4 Average track record is poor Well-known empirical studies documenting unprofitability of average market-timing strategy Authors (Year) Focus group that FAILS to successfully time the market on average Treynor and Mazuy (1966) Hendricksson and Merton (1981) Elton et al (1995) Becker et al (1999, 2001) Coggin and Hunter (1993) Barber and Odean (2000) Graham and Harvey (1997) Chance and Hemler (2001) Brooks and Gray (2004) Goyal and Welch (2005) Mutual funds Mutual funds Bond managers Asset allocation funds Pension funds Investment clubs Investment newsletters Professional market timers Economists Nearly all reported "predictive" variables > 4

5 Why does the average market-timer fail? Statistically locating turning points is backward looking, regardless of technique Historical required to identify a turning point NBER announces economic cycles with long delay Are you processing data in proprietary ways? Technical or extreme relative valuation models assume statistical process is deterministic In reality, relationships have stochastic components Even for a mean-reverting process, tracking error can be large and can erase alpha added from catching the turn Technical trading rules fail when codified in real-time Examples: Growth-value; LC-SC; US-international > 5

6 Why does the average market-timer fail? Complication: Economic data often revised Using real-time data often alters previous studies Federal Reserve of Philadelphia s real-time data library Perhaps no field of study in economics is potentially as sensitive as financial economics (Tom Stark, FRB-Philadelphia) Predictive variables are often not predictive in real time (Goyal and Welch, 2005) Variables that in real time do NOT predict future stock returns: div/p, div/v, e/p, book to market, default spread Difficult to find variables that beat the historical average (Campbell and Thompson, 2005 NBER WP) The one that works (term spread) may not in the future > 6

7 Inverted yield curves and future recessions Mixed signals out of sample Percentage (%) Recessions missed Recessions correct False positives Recessions correct Real GDP growth Yield curve (5-year Treasury minus Fed funds rate) > 7 Most empirical studies on yield curve conducted over shaded sample period Yield curve s TAA value depends critically on credibility regime

8 Real-time example of TAA strategy: Economic risk and duration tilts in Treasury bond portfolio Active bond management requires timing strategy No idiosyncratic risk / all systematic risk Unexpected inflation shocks key to sizable bond alpha Betting on the future path of inflation Concerns of rising interest rates abound Mean reversion: A persuasive and pervasive notion Reading the inflation tea leaves Active duration management Duration is the primary factor in tactical bond management Duration accounts for 86% of Treasury bond total returns > 8

9 How to anticipate future inflation changes before the market does Successful interest-rate forecasting requires superior inflation forecasting Potential: Inflation is a persistent process (autoregressive) Closely followed data believed to be leading indicators Efficient markets should price in publicly available data However, indicators may flow through to actual CPI inflation with a long and dynamic lag Markets may fail to optimally combine these signals If investors believe that they possess superior inflation forecasting ability, then they should pursue active duration management > 9

10 Forward-looking investment decision rules: Projected inflation versus market expectations % Confidence interval Tactical bet: shorten duration Tactical bet: lengthen duration Projected inflation path Market Inflation expectations (Embedded in term structure) t 0 t1 t2 t3 t4 time Tactical investor shortens portfolio s duration from t 1 to t 2 : Tactical investor lengthens portfolio s duration from t 3 to t 4 : At all other times, tactical investor owns broad duration exposure ( neutral to the benchmark ) Bet on higher-than-expected inflation Bet on lower-than-expected inflation > 10

11 Dynamic tactical duration strategies Dynamic forward-looking duration trading strategies since 1978 Real-time VAR model: Medium signal-band width criteria (0.5% std error) Tactical strategy performance (in %) Forecast horizon Transaction costs Mean monthly ER (alpha) Alpha s.d. Turnover rate (%) 1 month ahead Zero Low (0.25%) High (0.50%) 0.02) (0.03) (0.11) months ahead Zero Low (0.25%) High (0.50%) 0.01) (0.04) (0.09) months ahead Zero Low (0.25%) High (0.50%) 0.03) (0.02) (0.08) Excluding transaction costs, excess return is positive, but statistically insignificant. Transaction costs lead to significant portfolio loss (timing is losing strategy). In real time, leading indicator generates false signals and high turnover rates. Tighter signal-band width (e.g., 0.25%) leads to greater tactical losses. Turnover rate rises by more than the alpha content of the forecasting model. > 11

12 Theory, intuition, and conventional wisdom can all be wrong in a tactical strategy Link between commodity prices and inflation Empirical link varies over time and by sampling method appears strong, but CRB Commodity Price Futures Index CPI Inflation Rate (%) R^2 on dynamic OLS model: Full history R^2 on dynamic OLS model: Rolling window Even sophisticated forward-looking models can fail Empirical link has changed fundamentally and is now tenuous Regression-based models are always late to pick up such change Heisenberg s Uncertainty Principle > 12

13 Key takeaways for market-timing strategies Forecasting systematic risk factors is extremely difficult Unknown probabilities and magnitudes of the risk scenarios Regime changes: History / model may no longer work Theory, intuition, and conventional wisdom can all be wrong Systematic risk makes a powerful case for indexing and SAA decision Market-timing success is possible, but not probable Incur active risk in proportion to your information Rigorously test strategies out-of-sample, using real-time data & rules Market timing can quickly engender performance chasing How persistent is an active manager s timing alpha? Security selection offers more breadth than timing strategies Average active manager is more successful at security selection than at market timing > 13

14 Questions? 2005 The Vanguard Group, Inc. All rights reserved. Vanguard Marketing Corporation, Distributor. > 14

15 Appendices > 15

16 The Complementary Role for Tactical Asset Allocation: Value at the Margin % Wilshire 5000 TAA Strategy 60% / 40% Benchmark Monthly Mean Return Monthly Standard Deviation Vanguard Tactical Asset Allocation Strategy Monthly Excess Return Benchmark: 60% Wilshire 5000 / 40% Lehmann Long Term Treasury 6 4 Monthly Excess Return Jan-96 Jan-97 Jan-98 Jan-99 Jan-00 Jan-01 Jan-02 Jan-03 Jan-04 Jan-05 Monthly Strategy Excess Return 12 Month MA Equity Risk Premium > 16 Source: The Vanguard Group Mean Excess Return T-Statistic

17 Understanding Why Strategic Asset Allocation is Critical for Long Term Performance The Intuition: An index implementation of SAA (i.e., policy return ) outperforms with lower volatility The Statistics Supporting the Intuition: Results for Net Returns of 214 US Balanced Funds (12: :2003) Policy Return as % of Actual Return Policy Volatility as % of Actual Return % of Actual Return Variation Explained by Policy Return Variation Average % 90.23% 81.43% Median % 92.15% 85.29% Source: The Vanguard Group > 17