can fund managers asset allocate? andrew clare, dirk nitzsche & meadhbh sherman centre for asset...
TRANSCRIPT
Can fund managers asset allocate?
Andrew Clare, Dirk Nitzsche
& Meadhbh Sherman
Centre for Asset Management Research,Cass Business School, London.
Click to edit Master title style
Copyright: Andrew Clare, 2012 Page 2
Overview
• What we are assessing are TAA skills
• Previous, related work
• Data & methodologies
• Sub-set of results
• Summary
• All the results are preliminary; final version of paper should be available in September; also in process of updating the results
Click to edit Master title style
Copyright: Andrew Clare, 2012 Page 3
Some previous work in this area
• Sharpe (1988, 1992) asserted that:• a fund’s asset allocation decisions account for almost all of its fund’s
performance
• Brinson, Hood and Beebower (1986) examine the performance of 91 US pension funds using data from 1974 to 1983 they found that:• the policy mix explained 93.6 percent of the average fund’s return variation
over time (as measured by the R2)
• Brinson, Singer and Beebower (1991) quarterly returns data on 82 US pension funds spanning the period 1977 to 1987• active investment decisions did little to improve portfolio performance
because any abnormal performance was insignificant
Click to edit Master title style
Copyright: Andrew Clare, 2012 Page 4
Some previous work in this area
• Using UK pension fund data, Blake, Lehmann & Timmerman (1999)• the majority of return is derived from the strategic asset allocation decision
• Using data on large Canadian and US pension funds Andonov et al (2011) find:
• that changes in asset allocation, market timing and security selection generate positive abnormal returns of 17, 27 and 45 basis points per year respectively
• Using a small sample of US managers Weigel (1991) finds that
• the vast majority (over 75%) of managers exhibit positive, significant market timing ability
• managers that are good at market timing are paying for this skill in the form of negative returns to non-market-timing strategies
Click to edit Master title style
Copyright: Andrew Clare, 2012 Page 5
Data
• We collected monthly, net of fee returns data on multi-asset class retail funds managed in Canada, the UK and the USA• Data sample is January 2000 to December 2010 – 714 funds
• We also collected data on monthly proportions of multi asset class funds invested in: Cash, Govt bonds, Corporate bonds & Equities• Data sample is January 2006 to December 2010 – 355 funds
• We use the two data sets as the basis for two approaches to the problem
Click to edit Master title style
Copyright: Andrew Clare, 2012 Page 6
The multi-asset class funds
US All categoriesConservative
Allocation Moderate AllocationAggressive Allocation
Equity 50.17 30.98 58.57 68.08
Bond 34.99 48.61 29.88 17.23
Cash 9.54 12.35 7.74 6.04
Other 3.78 6.33 2.51 4.67
No. of funds 349 116 196 37
UK All categories Cautious Managed Balanced Managed Active Managed
Equity 63.76 45.24 68.74 80.97
Bond 17.17 33.20 12.99 2.31
Cash 10.15 12.78 9.80 7.23
Other 8.58 8.34 8.25 9.26
No. of funds 134 48 49 37
Canada All categoriesFixed Income
Balanced Neutral Balanced Equity BalancedTactical
BalancedEquity 48.48 26.01 53.26 64.1 64.05Bond 36.97 50.76 34.23 23.67 26.45Cash 10.53 14.62 10.06 7.08 8.53Other 1.26 1.66 1.06 1.05 0.73No. of funds 231 73 95 58 5
Click to edit Master title style
Copyright: Andrew Clare, 2012 Page 7
Methodology
• We apply variants of two methodologies to determine whether fund managers can ‘time’ their asset allocation decisions
• We effectively test for tactical rather than strategic timing abilities
• Methodology 1:
• This is based on the “conditional beta” approach which imputes timing ability using fund returns (Ferson and Schadt (1996)):
• We use several variants of this approach on the longer data set
Click to edit Master title style
Copyright: Andrew Clare, 2012 Page 8
Methodology 1
• If θ2 is positive it implies successful timing of equity market
• If θ3 is positive it implies successful timing of corp bond market
• If θ4 is positive it implies successful timing of govt bond market
• If θ5 is positive it implies successful timing of cash
Click to edit Master title style
Copyright: Andrew Clare, 2012 Page 9
Results – US and UK
1 2 3 4US All US- Conservative Allocation US- Moderate Allocation US- Aggressive Allocation
Timing Coeff % Sig
Pos% Sig Neg % Pos
% Neg%
Sig Pos
% Sig Neg % Pos
% Neg% Sig Pos
% Sig Neg % Pos
% Neg%
Sig Pos
% Sig Neg % Pos
% Neg
3 Equity 1.43 6.30 25.50 74.50 0.00 13.51 18.92 81.08 0.86 6.90 18.10 81.90 2.04 4.59 31.12 68.88
4 Corp 4.58 0.86 78.51 21.49 13.51 2.70 75.68 24.32 1.72 0.00 81.90 18.10 4.59 1.02 77.04 22.96
5 Govt 0.86 10.60 24.07 75.93 2.70 10.81 35.14 64.86 0.00 7.76 22.41 77.59 1.02 12.24 22.96 77.04
6 Cash 7.74 47.85 21.49 78.51 8.11 37.84 24.32 75.68 7.76 51.72 18.10 81.90 7.65 47.45 22.96 77.04
UK-AllUK- Cautious Managed UK- Balanced Managed UK- Active Managed
% SigPos
% Sig Neg % Pos % Neg
% Sig Pos
% Sig Neg % Pos % Neg
% Sig Pos
% Sig Neg % Pos % Neg
% Sig Pos
% Sig Neg % Pos % Neg
9 Equity 0 7.46 13.43 86.57 0.00 12.50 16.67 83.33 0.00 4.08 8.16 91.84 0.00 5.41 16.22 83.78
10 Corp 5.22 1.49 55.97 44.03 4.17 0.00 58.33 41.67 6.12 2.04 53.06 46.94 5.41 2.70 56.76 43.24
11 Govt 0 14.18 11.19 88.81 0.00 14.58 27.08 72.92 0.00 18.37 0.00 100.00 0.00 8.11 5.41 94.59
12 Cash 1.49 13.43 32.09 67.91 0.00 29.17 2.08 97.92 2.04 8.16 44.90 55.10 2.70 0.00 54.05 45.95
Timing coefficients
• Corporate bond timing more prevalent than equity market timing
• UK Cautious Managed, can’t seem to time cash!
Click to edit Master title style
Copyright: Andrew Clare, 2012 Page 10
Results – Canada
Canada-AllCanada- Fixed Income
BalancedCanada- Neutral
Balanced
Canada- Equity Balanced
Canada- Tactical Balanced
%SigPos
% Sig Neg
% Pos
% Neg
% Sig Pos
% Sig Neg
% Pos
% Neg
% Sig Pos
% Sig Neg
% Pos
% Neg
%Sig Pos
% Sig Neg
% Pos
% Neg
%Sig Pos
% Sig Neg
% Pos
% Neg
15 Equity 0.43 12.12 26.41 73.59 1.37 21.92 8.22 91.78 0.00 10.53 30.53 69.47 0.00 3.45 43.10 56.90 0.00 0.00 20.00 80.00
16 Cor Bond 3.46 3.46 43.29 56.71 0.00 6.85 16.44 83.56 5.26 1.05 52.63 47.37 5.17 3.45 60.34 39.66 0.00 0.00 60.00 40.00
17 Gov Bond 2.6 0.87 55.84 44.16 2.74 1.37 75.34 24.66 2.11 1.05 49.47 50.53 3.45 0.00 43.10 56.90 0.00 0.00 40.00 60.00
18 Cash 4.33 9.96 42.86 57.14 1.37 6.85 34.25 65.75 6.32 9.47 48.42 51.58 5.17 15.52 44.83 55.17 0.00 0.00 40.00 60.00
Timing coefficients
• Bond timing more prevalent than equity market timing for Canadian funds
• However, overall proportion that are found to have significant timing ability in all three markets is very low.
Click to edit Master title style
Copyright: Andrew Clare, 2012 Page 11
Results – summary
• Arguably Canadian managers are the better tactical asset allocators
US (%) UK (%) CAN (%)
% with All 4 classes Pos 0.29 0.00 1.30
% with 3 out of 4 Pos 7.45 4.48 9.96
% with 2 out of 4 Pos 35.24 23.13 46.75
% with 1 out of 4 Pos 55.59 52.99 38.96
% with none Pos 1.43 19.40 2.16
% with All 4 classes Sig + Pos 0.00 0.00 0.00
% with 3 out of 4 Sig + Pos 0.00 0.00 0.00
% with 2 out of 4 Sig + Pos 0.57 0.00 0.87
% with 1 out of 4 Sig + Pos 13.47 6.72 9.09
% with none Sig + Pos 85.96 93.28 89.18
Click to edit Master title style
Copyright: Andrew Clare, 2012 Page 12
Methodology 2
• This approach is much simpler and much more direct, than the returns-based approach
• It asks whether weightings rise/fall in proportion to market returns
• A positive value for β indicates that it does
• Again, we use a number of variants of this approach
%ACj,t = α + βjRj,t+1) + εj,t
Click to edit Master title style
Copyright: Andrew Clare, 2012 Page 13
Timing the equity market
60.2857.34
37.50
71.15
12.96
6.29 5.36
21.79
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
80.0
ALL US UK CAN
Pro
po
rtio
n o
f fu
nd
ma
na
ge
rs
Click to edit Master title style
Copyright: Andrew Clare, 2012 Page 14
Timing credit
36.06
24.48
28.57
49.36
4.51
1.4
5.367.05
0.0
10.0
20.0
30.0
40.0
50.0
60.0
ALL US UK CAN
Pro
po
rtio
n o
f fu
nd
ma
na
ge
rs
Click to edit Master title style
Copyright: Andrew Clare, 2012 Page 15
Timing the govt bond market
52.68
66.43
46.43
42.31
5.63 6.29
16.07
1.28
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
ALL US UK CAN
Pro
po
rtio
n o
f fu
nd
ma
na
ge
rs
Click to edit Master title style
Copyright: Andrew Clare, 2012 Page 16
Timing cash
55.49 55.94
37.50
61.54
29.58
34.27
3.57
34.62
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
ALL US UK CAN
Pro
po
rtio
n o
f fu
nd
ma
na
ge
rs
Click to edit Master title style
Copyright: Andrew Clare, 2012 Page 17
DGF … the new balanced
• Most of the research says that strategic asset allocation gives the biggest bang for one’s buck
• In this work we are really looking at TAA
• There have been a huge number of DGFs launched recently; there is SAA embedded in these funds
• Some DGFs emphasise the TAA overlay as an added source of return
Click to edit Master title style
Copyright: Andrew Clare, 2012 Page 18
Summary
• These are just a small set of the preliminary results
• But the basic finding is that:• Using returns-based data there is little evidence of TAA ability amongst
these managers
• Using asset class weights, there is much more evidence of TAA ability
• However, in both cases it is still very difficult to distinguish this skill from luck
• But as we know, sometimes it’s better to be lucky than good!