Buy and Sell Timing Decisions by Mutual Fund Managers

√
Buy and Sell Timing Decisions
by Mutual Fund Managers
Rajiv D. Banker
Janice Chen
Fox School of Business
Temple University
Theme of the Paper √
• Job security and compensation incentives imply
that active fund managers have superior ability
to time the market in buy or sell trades
• Psychological biases, such as disposition or
endowment effects, may inhibit efficient timing
• Data on 18 fund managers at one firm spanning
one year (2005)
• We find fund managers exhibit good buy and sell
timing ability
• Actual trading strategy outperforms both
momentum and contrarian strategies
Mutual Funds
• A mutual fund is a trust that pools the
savings of a number of investors
• In end 2007, the combined assets of
mutual funds in the U.S. were $12 trillion
• In early 2008, the worldwide value of all
mutual funds totaled more than $26 trillion
• Mutual fund companies competed
vigorously to attract investor funds
Fund Manager Performance
Stock
Picking Ability
Fama
(1972)
Market
Timing Ability
Wermers (1997)
Jiang et al. (2007)
Grinblatt and Titman
(1989, 1993)
Bollen and Busse (2001)
Principal Hypothesis √
Fund managers have the ability to
time their buy and sell trades
efficiently
Timing ability is critical for fund managers:
• High liquidity and frequent trading activities
• Replacement of less successful fund managers
• High investment turnover rate
• Bonuses based on fund performance
Measures of Timing Ability √
Based on only fund-level data
• Treynor and Mazuy (1966)
R p   p   p * Rm  c p * Rm   p
2
• Henriksson and Merton (1981)
Rp   p   p * Rm  c p * max( 0, R m )   p
• Jiang et al. (2007)
Artificial timing bias: time-varying parameter problem
E (rpt 1 )  E (  pt ) E (rmt 1 )  cov(  pt , rmt 1 )
√
Behavioral Finance Studies
• Psychological biases inhibit performance
• Overconfidence explains reduction in
investors’ performance due to excessive
trading (Barber and Odean 2000, 2001,
2002)
• Fund managers exhibit herding behavior
(Grinblatt et al. 1995)
• Disposition effect (Shefrin and Statman 1985)
• Endowment effect (Thaler, 1980)
Prospect Theory
Kahneman and Tversky 1979
VALUE
LOSSES
Disposition Effect
• Disposition effect is an implication of
prospect theory (Kahneman and Tversky
1979) applied to investment decisions
• Disposition effect reflects the tendency of
investors to sell winners too early and hold
on to losers too long (Shefrin and Statman
1985)
Disposition Effects on Trading √
• Individual investors sell winners early and
hold on to losers (Odean 1998)
• Professional managers also exhibit
disposition bias (Grinblatt and Keloharju
2001, Scherbina and Jin 2005)
• Disposition effect may lead to
inefficient timing decisions for buy
and sell trades
Endowment Effect
√
• People often demand much more to give
up an object
• Tversky and Kahneman (1991) attribute
this phenomenon to loss aversion
(tendency to strongly prefer avoiding
losses than acquiring gains)
• Endowment effect may lead to
inefficient timing decisions for buy
and sell trades
Alternative Trading Strategies
• Contrarian Strategy
Buying losers and selling winners (De Bondt and
Thaler 1985, 1987)
• Momentum Strategy
Buying winners and selling losers (Jegadeesh
and Titman, 1993)
• If fund managers are good at timing their
trades, they should outperform these
simple strategies
√
Hypothesis 2
• The actual trading strategy of fund
managers outperforms both the contrarian
strategy and the momentum strategy
trading the same stocks
Sample Data
√
• Transaction-based datasets
• 11,906 buy transactions
8,465 sell transactions
• Each transaction has trading volume
(quantity), market value, cumulative
excess returns (3, 6 month pre-trade
returns and 3, 6 month post-trade returns).
• There are 18 fund managers
• Our transaction dataset is proprietary and
spans one calendar year 2005
√
Four Time Periods for Annualized
Excess Returns
All 4 time periods are calibrated relative to the date of trade
Types of Traded Stocks
Buy Transactions
Sell Transactions
Number
5,399
%
45%
Number
4,019
%
47%
Value
Large cap
Small cap
Over-weighted
6,203
7,929
3,673
10,194
52%
67%
31%
86%
4,171
5,483
2,707
6,050
49%
65%
32%
71%
Under-weighted
1,712
14%
2,402
28%
Growth
Empirical Results
• Buy sample:
The medians of annualized excess returns are
significantly negative in pre-trade periods while
significantly positive in post-trade periods
• Sell sample:
The medians of annualized excess returns are
significantly positive in pre-trade periods while
significantly negative in a post-trade period
Buy Transactions
Median Excess Returns
2
1.44
1
median excess returns
√
0.31
0
-1
-6 -6
-3
-3
0
+3
-2
-3
-2.17
-4
-5
-6
-7
-5.90
+3
+6
Sell Transactions
√
Median Excess Returns
10
8.78
median excess returns
8
6
4.05
4
1.48
2
-6
0
-6
-2
-4
-3
0
+3
-3
-1.98
+6
Buy Sub-Sample Analysis √
Median Excess Returns
Pre6-3
Pre3-0
Post0-3
Post3-6
Full Sample
-2.17**
-5.90***
1.44***
0.31***
Growth
0.39***
-0.35***
3.42***
-1.18***
Value
-4.18***
-9.44***
-0.13***
0.97***
Large Cap
-2.45
-5.83***
1.92***
-1.92
Small Cap
-0.34***
-4.87***
0.27***
4.37***
Overweighted
-2.54
-6.31***
1.56***
0.15***
Underweighted
0.02***
-3.54
0.59***
1.19***
√
Sell Sub-Sample Analysis
Median Excess Returns
Pre6-3
Pre3-0
Full Sample
4.05***
8.78***
Growth
10.94***
14.76***
Value
-0.26***
Large Cap
Post0-3
Post3-6
-1.98***
1.48***
-2.66
2.86***
2.52***
-1.26***
0.29***
3.64***
6.08***
-2.25**
0.19***
Small Cap
5.87***
15.89***
-1.31***
4.14***
Overweighted
4.78***
12.27***
-2.15***
2.43***
Underweighted
2.20***
2.19***
-1.41
-0.54***
√
Paired Tests
Pre-Trade Minus Post-Trade Median Excess Returns
Buy Sample
Sell Sample
Post0-3
Post3-6
Post0-3
Post3-6
Pre3-0
-5.23***
-4.53***
9.67***
7.63***
Pre6-3
-2.49***
-1.99***
5.55***
2.70***
Buy Sample Paired Tests
Pre-Trade Minus Post-Trade Median Excess Returns
Buy
0
-1
Post0-3
Post0-3
-2
-3
-4
-5
-6
Pre3-0
Pre6-3
Sell Sample Paired Tests
Pre-Trade Minus Post-Trade Median Excess Returns
sell
12
10
8
Pre3-0
6
Pre6-3
4
2
0
Post0-3
Post0-3
Growth vs. Value
Paired Test
Buy Sample
Growth
Post0-3
Post3-6
Sell Sample
Post0-3
Post3-6
Pre3-0
-0.82***
2.90***
16.00***
12.06***
Pre6-3
-1.82
3.96***
12.99***
7.69***
Buy Sample
Value
Post0-3
Post3-6
Sell Sample
Post0-3
Pre3-0
-8.63***
-9.99***
5.26***
Pre6-3
-2.57***
-5.46***
-0.54
Post3-6
1.71***
-0.82
Large Cap vs. Small Cap
Paired Test
Buy Sample
Large Cap
Post0-3
Pre3-0
-6.57***
Pre6-3
-2.50***
Sell Sample
Post3-6
-1.87***
0.15
Buy Sample
Small Cap
Post0-3
Post0-3
Post3-6
7.19***
4.90***
4.68***
3.09***
Sell Sample
Post3-6
Post0-3
Post3-6
Pre3-0
-1.33**
-9.41***
17.02***
12.82***
Pre6-3
-1.71
-5.36***
7.93***
2.70***
Overweighted vs. Underweighted
Paired Test
Buy Sample
Overweighted
Post0-3
Sell Sample
Post3-6
Post0-3
Post3-6
Pre3-0
-5.88***
-4.77***
12.70***
10.84***
Pre6-3
-3.28***
-2.90***
6.89***
2.71***
Buy Sample
Underweighted
Sell Sample
Post0-3
Post3-6
Post0-3
Post3-6
Pre3-0
-1.03
-1.90
4.02***
1.27**
Pre6-3
1.64
2.48
2.47***
2.69*
√
Momentum and Contrarian Strategies
• Winners are stocks with positive excess
returns in 3 months preceding trade date
• Losers are stocks with negative excess
returns in 3 months preceding trade date
• Momentum strategy mimicking portfolios
buy winners and sell losers
• Contrarian strategy mimicking portfolios
buy losers and sell winners
√
Comparison with Momentum Strategy
Excess returns of actual strategy – Excess returns of momentum strategy
Mean
Diff_Ret
Buy Sample
Sell Sample
Mean
1.53
-2.13
Median
p-value
0.03
0.03
Median
1.57
-2.16
p-value
0.015
0.001
√
Comparison with Contrarian Strategy
Excess returns of actual strategy – Excess returns of contrarian strategy
Mean
Diff_Ret
Buy Sample
Sell Sample
Mean
1.37
-1.97
Median
p-value
0.075
0.01
Median
1.26
-1.85
p-value
0.03
0.003
Robustness Check √
• Results survive when performance of
trading strategies is evaluated relative to
distribution of simulated excess returns
(Kothari and Warner, 2001) drawn from the
CRSP population of stocks
Conclusion
√
• Managers have a good market timing ability to
buy stocks at a low price and sell at a high price
• Fund managers are not overly influenced by
psychological biases such as the endowment
effect or the disposition effect
• Fund managers outperform both the momentum
strategy and the contrarian strategy
• Caveat: Data for only one company for one year