factor efficiency - hsbc · volatility times correlation, msci barra research. 4 ... for a...
TRANSCRIPT
2 | HSBC Global Asset Management
Factor EfficiencyExecutive Summary
This note sets out the premise behind our views on
efficient beta investing and describes a method of
identifying the most ‘efficient’ factor driven strategies.
Smart beta indexes and Exchange Traded Funds (ETFs)
have become increasingly popular, with many providers
offering a broad suite of such strategies. Most are
based on well recognised risk premium factors such as
value, small cap, momentum, low volatility and quality.
However, many such indexes have unintended risks
and additional exposure to untargeted factors because
of their construction methodology. We investigate a
method of determining the efficiency of a factor index
and apply it to our smart beta investment solutions. By
showing that pure indexes have larger factor efficiency,
we can conclude that they are relatively efficient: they
have a larger portion of active risk attributed to the
targeted factor.
HSBC Global Asset Management | 3
IntroductionThe appeal behind smart beta indexes is that they are
systematic and transparent, rendering them easy to construct
and rebalance. Smart beta indexes are also a low cost way
for an investor to obtain exposure to factors which they might
be lacking within their portfolios. Many smart beta indexes
tend to be constructed with an emphasis on simplicity,
often using simple sorting and weighting techniques based
on either a single factor (e.g. book-to-price) or a composite
score (e.g. value). Some smart beta indexes are constructed
to maximise investability, combining factor tilts with market
cap weighting. Unfortunately, even though this sort of
construction technique tends to result in higher factor
exposures, there are usually few restrictions on other style
characteristics. This leads to unintended factor exposures
and thus undesired risk.
The inspiration for this paper comes from a recently
published paper by Hunstad and Dekhayser1, where they
introduce a new measure called the factor efficiency ratio.
Much of the debate on smart beta index construction stems
from unintended factor exposures. These are usually the
result of conforming to the requirements of transparency,
simplicity and investability. Indexes are constructed from
stocks ranked by factor of interest (e.g. factor indexes
sold by FTSE, MSCI and Russell). There is evidence that
simple minimum variance optimisation, a common smart
beta strategy, results in time-varying factor exposures2.
Goldberg et al. suggest that it is important to be aware of
these exposures and highlight the benefits of targeting pure
exposures when building such indexes.
In what follows we describe how the factor efficiency
ratio is calculated, how it should be interpreted and
relevant applications.
1 Hunstad and Dekhayser (2014) – Evaluating the Efficiency of ‘Smart Beta’ Indexes, Working Paper.
2 Goldberg, Leshem and Geddes (2013) – Restoring Value to Minimum Variance, Forthcoming in Jour-nal of Investment Management Davis and Menchero (2010) – Risk Contribution is Exposure times Volatility times Correlation, MSCI Barra Research.
4 | HSBC Global Asset Management
MethodologyThe factor efficiency ratio is defined as the ratio of tracking
error from the desired factor(s) to the total tracking error.
It follows that this can be used to measure the efficiency
of an index, i.e. the ratio of desired to undesired active risk.
Formally, it is constructed as follows:
Factor Efficiency Ratio =∑ARD
AR - ∑ARD
∑ARD is the sum of active risk contributions of the desired
factors while AR is the total active risk of the portfolio. Here
we look at active risk contribution as opposed to absolute risk
contribution: the purpose of index is to add exposure beyond
that of a specified benchmark (i.e. we look at active style
exposures). For a single-factor smart beta value index,
the factor efficiency ratio is calculated as follows:
Factor Efficiency RatioValue =∑ARValue
AR - ∑ARValue
Ex-ante active risk can be decomposed as follows
(this is discussed in detail in Davis, Menchero3:
AR = Var(RA) = ∑x σFi ρFi,R
A + ∑w σuk ρuk,RA
where σFValueρFValue RA is otherwise equivalent to the
marginal contribution to risk of the value factor.
The factor efficiency ratio is therefore by construction
an ex-ante ratio which decomposes the risk the portfolio
expects to deliver into desirable and undesirable active risk.
iiA
KA
The variables above can be estimated using a factor risk
model. We use our internal factor risk model which is broadly
similar to the commercially available alternatives from Barra,
Axioma and Bloomberg. It is a cross-sectional risk model
that decomposes systematic market risk into 10 broad style
factors and industry, country and currency factors. The
cross-sectional variance-covariance matrix is built using
10 years of weekly data and covariance shrinkage.
DataWe first obtains the smart beta index constituents and
weights through time for the indexes, the MSCI style indexes
and the RAFI indexes. We care about changes over time as
the calculated factor efficiency is time-varying. Next, we
construct the factor risk model specific to the investment
universe, in this case the MSCI World. We extract factor
exposures and factor returns from the risk model, which
we use to calculate marginal active contributions to risk.
Finally, we calculate the factor efficiency ratio for each of
pure indexes (value, momentum, small cap, low volatility and
quality) and compare them to their closest competitors.
Total Active Risk
Sysrematic Active Risk
SystematicActive Risk
IndustryGroup
Style
Country
Currency
Total Active Risk Specific
Active Risk
Figure 1: Decomposition of total active risk in cross-sectional factor risk model
3 Davis and Menchero (2010) – Risk Contribution is Exposure times Volatility times Correlation, MSCI Barra Research.
i
HSBC Global Asset Management | 5
Results
As of March 2015
Hunstad and Dekhayser highlight the disparity between
active exposure and factor efficiency. An index with high
exposure to a particular factor will not necessarily have high
factor efficiency. For example, a simple ranking of value
stocks often has significant small-cap exposure. If we were
to buy the top quintile of value names, we would anticipate a
high exposure to both value and small cap factors. We would
prefer an efficient beta index to have a large proportion of
active risk driven by the targeted style factor and minimal
active risk contributions from the other factors.
From the results we can demonstrate that pure indexes
are indeed relatively ‘efficient’ from a factor efficiency ratio
perspective. This is not surprising as HSBC’s index construction
process aims to minimise unwanted factor exposures as well as
broad active risk due to its active weight constraints.
In Figure 2 we show the factor efficiency ratios for pure indexes.
Figure 2: Factor efficiency ratios Factor Efficiency Ratios as of March 2015
0.0
0.3
0.6
0.9
1.2
1.5
Momentum Quality Small Cap
Value Low Volatility
As can be seen from Figure 2, high factor exposure does
not directly correlate with factor efficiency. For example, the
momentum index has the lowest factor exposure at 0.47 but
a high factor efficiency ratio of 1.40. This is understandable
as momentum is a more volatile factor with a higher active
risk contribution. Looking at the desired factor’s exposures
alone might be misleading. Factor exposures fail to take
into account the risks from exposure to other potentially
undesirable factors.
Concentrating on active risk contribution also connects back
to the general debate on risk premia factors. There is a degree
of risk in investing in factors and their returns are time-varying.
Note that indices with the same factor exposures may
have different active risks based on the nature of the factor.
An index that is factor efficient has less contribution from
undesired risks. The key point is that we are only taking a risk
on the factors that we choose to invest in.
Total Active Risk (%)Factor Contribution to Active Risk (%)
Proportion of Factor Contribution to Active Risk (%)
Factor Efficiency Ratio
Exposure
Momentum 2.15 1.25 58.4 1.40 0.47
Quality 1.42 0.19 13.1 0.15 0.76
Small Cap 3.38 0.85 25.1 0.34 1.37
Value 2.3 0.54 23.3 0.30 0.91
Low Volatility 2.47 0.99 40.2 0.67 0.45
Source (for both the table and chart): Factset, Worldscope, Thomson Reuters, MSCI.
6 | HSBC Global Asset Management
Figure 3 is the decomposition of active risk for value pure index as of March 2015. This is a common risk attribution output from
portfolio attribution packages. As is evident from both the factor efficiency ratio and the breakdown (Figure 3), the active risk
contribution from value is significantly larger than most other active risk components. This is indeed what we expect to see from
an “efficient” index.
Figure 3: Decomposition of total active risk (31 March 2015) Pure Value Enhanced Value
When comparing this to the MSCI World Enhanced Value Index (MSCI’s high exposure version of a value index), we can see how
different the index is in terms of factor efficiency ratios. The factor efficiency ratio of the MSCI World Enhanced Value Index at
the same point in time is 0.09, compared to figure of 0.30. This implies that index takes on approximately 5 times as much value-
related active risk than the MSCI World Enhanced Value Index per 1% of non-value active risk. At the same time, it has a similar
active exposure to value as the MSCI World Enhanced Value (High Exposure) index (0.91 vs 0.84). Thus, in a sense, the Pure
Value index manages to provide equally high exposure to the value factor coupled with lower active risk from other exposures.
Looking at the decomposition of style active risk for the MSCI index on Figure 4, we can see that even though the biggest
component of active risk is indeed value, there are still significant active risk contributions from size, growth and momentum.
A significant portion of its active risk comes from country active risk. A closer look at the breakdown shows that the majority
of this comes from active exposure to Japan. This does not appear to be consistent with an “index” strategy.
Figure 4: Decomposition of style active risk Pure Value Enhanced Value
0%
20%
40%
60%
80%
100%
10%
30%
50%
70%
90%
Style Countries Industries Currencies Non-Factor0%
20%
40%
60%
80%
100%
10%
30%
50%
70%
90%
Style Countries Industries Currencies Non-Factor
Contribution Positive Contribution Negative Total
0%
5%
10%
15%
20%
25%
Volatility
Value
Trading Activity
Size
Profitability
Mom
entum
Leverage
Grow
th
Earnings Variability
Dividend Yield
Volatility
Value
Trading Activity
Size
Profitability
Mom
entum
Leverage
Grow
th
Earnings Variability
Dividend Yield
0%
4%
8%
12%
16%
18%
Contribution Positive Contribution Negative
Source: Factset, Worldscope, Thomson Reuters, MSCI.
Source: Factset, Worldscope, Thomson Reuters, MSCI.
HSBC Global Asset Management | 7
To extend the comparison between the funds with other style indexes, we look at the two ranges of MSCI factor indexes:
their high exposure range as well as their high capacity range. The latter impose tilts on the cap-weighted index to increase
investability. In terms of factor efficiency ratios, as of the end of March 2015 the pure indexes appear to be more ‘efficient’
than both ranges. They also exhibit consistently higher exposures than the tilt range.
Figure 5: Comparison of factor efficiency ratios Factor Efficiency Ratios as of March 2015
Figure 6: Comparison of active exposures Active Exposures as of March 2015
MSCI High Exposure
Total Active Risk (%)
Factor Contribution to Active Risk (%)
Proportion of Factor Contribution to Active Risk (%)
Factor Efficiency Ratio
Exposure
MSCI Momentum 4.01 1.12 27.9 0.39 0.46
MSCI Quality 2.9 0.10 3.4 0.04 1.02
MSCI Equal Weighted 2.52 0.33 13.1 0.15 0.94
MSCI Enhanced Value 3.69 0.30 8.2 0.09 0.84
MSCI Minimum Vol 3.94 0.78 19.7 0.25 0.33
0.0
0.3
0.6
0.9
1.2
1.5
Momentum Quality Small Cap Value Low Volatility
HSBC’s Pure Factors MSCI Tilt MSCI High Exposure
0.0
0.3
0.6
0.9
1.2
1.5
HSBC’s Pure Factors MSCI Tilt MSCI High Exposure
QualityMomentum Small Cap Value Low Volatility
Source: Factset, Worldscope, Thomson Reuters, MSCI.
Source: Factset, Worldscope, Thomson Reuters, MSCI.
8 | HSBC Global Asset Management
Figure 7: 12 month rolling factor efficiency ratios
Note that momentum and low volatility have the highest
variability in factor efficiency ratios. Ex-ante active risk is a
product of exposure, factor volatility and correlations, so any
of these can drive fluctuations in active risk composition.
Quality has the lowest average factor efficiency ratio through
time. This suggests that there are other active risks we
might wish to consider when constructing a quality index.
Quality might be harder to capture as a pure factor because
it is naturally skewed in terms of exposures to countries and
industries. We examined this by looking at snapshots of the
quality index active risk decomposition at different points in
time. A substantial part of active risk appears to come from
countries in both 2009 and 2015.
Time seriesNext we examine the changes in factor efficiency over
time using indexes, for which we have full constituent
and weighting data. We are unable to do the same for our
competitors’ indexes due to the lack of data. Instead we look
at annual snapshots of the indexes as far as they are available.
The charts in Figure 7 show that factor efficiency ratios
vary over time and are dependent on the active risk of the
targeted factor relative to the other factors. However, we
expect the factor efficiency ratio of pure indexes to be higher
over time. This is because the evaluation of active risk is a
product of exposure and marginal contribution to risk. Pure
indexes explicitly target exposure to other style factors so
the risk contribution of the main factor of interest should be
proportionally higher.
0.0
0.5
1.0
1.5
2.0
0.0
0.2
0.4
0.6
11/201211/2010 11/201311/2011 11/2014
Fact
or E
ffici
ency
Rat
io
Momentum Quality
Fact
or E
ffici
ency
Rat
io
0.0
0.5
1.0
1.5
0.5
1.0
1.5
Fact
or E
ffici
ency
Rat
io
0.0
Small Cap Value
Fact
or E
ffici
ency
Rat
io
0.5
1.0
1.5
2.0
0.0
Low Volatility
Fact
or E
ffici
ency
Rat
io
11/201211/2010 11/201311/2011 11/2014
11/201211/2010 11/201311/2011 11/2014 11/201211/2010 11/201311/2011 11/2014 11/201211/2010 11/201311/2011 11/2014
0.0
0.5
1.0
1.5
2.0
0.0
0.2
0.4
0.6
11/201211/2010 11/201311/2011 11/2014
Fact
or E
ffici
ency
Rat
io
Momentum Quality
Fact
or E
ffici
ency
Rat
io
0.0
0.5
1.0
1.5
0.5
1.0
1.5
Fact
or E
ffici
ency
Rat
io
0.0
Small Cap Value
Fact
or E
ffici
ency
Rat
io
0.5
1.0
1.5
2.0
0.0
Low Volatility
Fact
or E
ffici
ency
Rat
io
11/201211/2010 11/201311/2011 11/2014
11/201211/2010 11/201311/2011 11/2014 11/201211/2010 11/201311/2011 11/2014 11/201211/2010 11/201311/2011 11/2014
Source: Factset, Worldscope, Thomson Reuters, MSCI: Nov 2010 to April 2015.
HSBC Global Asset Management | 9
Figure 8: Pure quality index active risk decomposition
Thus a potential improvement for quality indexes would be through the direct control of country and industry exposures.
However this would inevitably reduce exposure to quality as the additional constraints would shrink the feasible set of solutions.
Figures 9 and 10 chart the changes in factor efficiency ratios over time for different indexes. They imply that if the index
construction process does not change over time, factor efficiency ratios should stay within an expected range.
Figure 9: Comparison of factor efficiency ratios for value indexes
Figure 10: Comparison of factor efficiency ratios for quality indexes
0%
20%
40%
60%
80%
100%
10%
30%
50%
70%
90%
Style Countries Industries Currencies Non-Factor0%
20%
40%
60%
80%
100%
10%
30%
50%
70%
90%
Style Countries Industries Currencies Non-Factor
Contribution Positive Contribution Negative Total
0.0
0.2
0.4
0.6
0.8
1.0
2012 2013 2014 2015
AMG Pure Value MSCI Value Tilt
0.00
0.05
0.10
0.15
0.20
2012 2013 2014 2015
AMG Pure Quality MSCI World Quality
0.0
0.5
1.0
1.5
2.0
2.5
20122011 2013 2014 2015
AMG Pure Low Volatility MSCI World Minimum Volatility
0.0
0.2
0.4
0.6
0.8
1.0
2012 2013 2014 2015
AMG Pure Value MSCI Value Tilt
0.00
0.05
0.10
0.15
0.20
2012 2013 2014 2015
AMG Pure Quality MSCI World Quality
0.0
0.5
1.0
1.5
2.0
2.5
20122011 2013 2014 2015
AMG Pure Low Volatility MSCI World Minimum Volatility
0.0
0.2
0.4
0.6
0.8
1.0
2012 2013 2014 2015
AMG Pure Value MSCI Value Tilt
0.00
0.05
0.10
0.15
0.20
2012 2013 2014 2015
AMG Pure Quality MSCI World Quality
0.0
0.5
1.0
1.5
2.0
2.5
20122011 2013 2014 2015
AMG Pure Low Volatility MSCI World Minimum Volatility
Figure 11: Comparison of factor efficiency ratios for low volatility indexes
The difference of factor efficiency
ratios is the result based on the
analysis above, it is natural to conclude
that the pure indexes appear to be
more efficient than the series of MSCI
indexes. This is a result of the index
construction process minimising
exposures to other factors, indirectly
causing a decrease in active risk
contribution from styles.
Figures 9, 10 & 11 Source: Factset, IBES, Worldscope, MSCI.
Sources and date of risk decomposition: Factset, Worldscope, Thomson Reuters, MSCI, March 2015.
10 | HSBC Global Asset Management
There appears to be a positive relationship between the
factor efficiency ratios and the correlations of the factor
index with pure factor returns. As discussed above, quality
has the lowest factor efficiency ratio. It also has the smallest
resemblance to that of a pure quality index.
Factor efficiency ratio vs return correlationsA natural extension of the discussion above is to check
how the return series of pure indexes correlate with long-
short pure style portfolio returns (with unit exposure to the
targeted factor and neutrality to other factors) and how this
relates to factor efficiency ratios. Mathematically, pure style
portfolio returns f_kt are calculated from the cross-sectional
regressions, where X_nkt represents the factor exposure of
stock n to factor k at time t.
Note that pure indexes are long-only index strategies so are
not necessarily representative of pure factor returns.
Figure 12: Correlation with pure style returns
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
00 .2 0.40 .6 0.8
Correlation of Smart Beta Index with Pure Factor Returns
Low Volatility
Quality
Value
Momentum
Small Cap
k=1
KReturnn,t = ∑Xn,k,t fk,t + εn,t
Sources of data and date of risk attribution: Factset, Worldscope, Thomson Reuters, MSCI, July 2001 - March 2015,
HSBC Global Asset Management | 11
ConclusionFactor efficiency ratios are a simple way of
determining the efficiency of smart beta indexes.
By measuring the ratio of desirable active risk to
undesirable active risk, we can evaluate the efficiency
of an index. There have been concerns that smart beta
indexes actually capture unintended factor, country,
industry or currency risks. The factor efficiency ratio
is an effective way of quantifying relative intended
to unintended tracking error, allowing investors to
distinguish between indexes with undesirable or
perhaps unforeseen risks from those that are truly
“efficient”. Investors can then devote most of their
active risk budget towards the factor of interest.
This note highlights the importance of focusing
on criteria beyond simple factor exposure when
buying a smart beta index. More sophisticated index
construction techniques, although more complicated
than simple sorting and weighting, can significantly
improve the overall purity and efficiency of smart
beta strategies.
Important InformationFor Professional Clients only and should not be distributed to or relied upon by Retail Clients.
The material contained herein is for information only and does not constitute investment advice or a recommendation to any
reader of this material to buy or sell investments.
This document is not intended for distribution to or use by any person or entity in any jurisdiction or country where such
distribution or use would be contrary to law or regulation. This document is not and should not be construed as an offer to sell
or the solicitation of an offer to purchase or subscribe to any investment.
Any views expressed were held at the time of preparation and are subject to change without notice. While any forecast,
projection or target where provided is indicative only and not guaranteed in any way. HSBC Global Asset Management (UK)
Limited accepts no liability for any failure to meet such forecast, projection or target.
The value of investments and any income from them can go down as well as up and investors may not get back the amount
originally invested. Where overseas investments are held the rate of currency exchange may also cause the value of such
investments to fluctuate. Investments in emerging markets are by their nature higher risk and potentially more volatile than
those inherent in some established markets. Stock market investments should be viewed as a medium to long term investment
and should be held for at least five years. Any performance information shown refers to the past and should not be seen as an
indication of future returns. Some of the ETFs invest predominantly in one geographic area; therefore any decline in the economy
of this area may affect the prices and value of the underlying assets.
To help improve our service and in the interests of security we may record and/or monitor your communication with us. HSBC
Global Asset Management (UK) Limited provides information to Institutions, Professional Advisers and their clients on the
investment products and services of the HSBC Group.
Approved for issue in the UK by HSBC Global Asset Management (UK) Limited, who are authorised and regulated by the
Financial Conduct Authority. Copyright © HSBC Global Asset Management (UK) Limited 2015. All rights reserved.
27505CP/ED/1015/ FP15-1849 EXP 31/07/2016.