GLOBAL MARKETS RESEARCH Doubling up on Momentum Value strategy in rates Position unwinding in Japan Global Quantitative Research Monthly 15 April 2014 See Appendix A-1 for analyst certification, important disclosures and the status of non-US analysts.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
certification, important
non-US analysts.
Content Page
Chapter I Doubl ing up on momentum. This is not 2009 (or 2000)
4
The sharp style moves seen in Europe, the US and globally since 21
March have seen Momentum sell off by 7%, 5% and
6% respectively to the close of Monday 14 April. These are very
large moves. With Momentum having now given up all of
its gains since we went long the style in early January, our call
to be long Momentum in Europe and globally is losing
money.
As a result of the recent moves, the style is now cheaper
than it has ever been before. In Europe, the high-momentum
stocks are now trading at a 27% discount to the low-momentum stocks
– this is unprecedented. Globally, the ‘winner’
stocks are also trading at historical lows. Previous sustained
Momentum sell-offs have always happened from much
higher multiples.
We consider several explanations that we have heard proposed for
the recent moves, including the Nasdaq sell-off,
various macro triggers, an EM-driven rotation and fund delevering.
Although we think these have some merit, none
seems to offer a compelling or complete explanation, nor a reason
for these style trends to continue.
We see recent underperformance of Momentum as a buying opportunity
both in Europe and globally. We are doubling up
our position in Momentum in our Global Recommended Quant
Portfolio.
Al la Harmswo rth
Chapter II Of bonds and bubbles – value outperforms in rates, as in
equities 14
Value investing is nothing new. Especially in equities, value
investing is among the oldest and most well-known strategies.
As early as 1934, Benjamin Graham and David Dodd spelt out
the principles of value investing in their text 'Security
Analysis'. Academics such as Eugene Fama and Kenneth French
have also studied value formally. For example, Fama &
French (1992) used value as part of their three-factor model of
stock returns. Legendary investors like Warren Buffett and
Seth Klarman famously use the principles of value to guide their
investment decisions.
In this article, we focus on value in interest rates. W e find that
using techniques fundamentally similar to those used in
equities can help rates investors identify value across both curves
and currencies. We discuss the following points:
• Value exists in both equities and interest rates.
• Carry in interest rates is analogous to earnings yield (E/P) in
equities. Just as stocks with high earnings yield outperform
those with low earning yield, curve points with high carry
outperform those with low carry.
• Comparing earnings yield both relative to its own history
(time-series) and across stocks (cross-sectional) is important
to
place current valuations in context. Similarly, comparing carry on
both time-series and cross-sectional basis is key.
• A long/short value strategy in rates outperforms a long-only
benchmark. Being free to trade the best points of value
across curves and currencies, without constraining the strategy to
hold pre-specified pairs, can help make the strategy
robust.
• Bubbles develop gradually, be it in equities or rates. Equities
and rates do not sell off when they are cheap – crashes
occur in the context of stretched valuations. Value-signals can
help track a build-up in overvaluation and identify when
sell-
offs in rates are more likely.
• A long-only version of the value strategy outperforms a long-only
benchmark on a duration-matched basis. By going long
the cheapest curve points across countries, the strategy beats a
static benchmark that holds the entire market.
• A short-only version of the value strategy can help hedge rising
rates. It can also be cheaper than being the short the
entire market as it identifies the most expensive points to be
short rather than shorting all points.
• Value can outperform when Momentum suffers. Momentum tends to
underperform when rates are close to zero and
markets range-bound. In such trendless conditions, value or
mean-reversion trades tend to outperform.
Anthony Morr is
Chapter I II Concentrated posit ion unwinding on 24 March 28
From 24 March 2014 through the morning’s trading session on 25
March, actively managed Japanese equity funds faced
a relatively strong headwind, especially in large cap universes.
Strategies based mainly on analyst earnings forecast-
related factors and stocks with high active-fund ownership ratios
produced negative returns of less than -3% over this
period. It is highly probable that, for some reason, there was a
concentrated spate of selling for position unwinding
purposes by active funds on those two days. The sell-off appears to
have come to an end before the start of the afternoon
trading session on 25 March, but we think there is still a need for
caution.
We all remember the 'quant headwind' in August 2007, when selling
pressure was similarly concentrated on certain
investment styles, such as E/P-based factors, and the performance
of a number of factors plummeted at the same time in
multiple markets around the world. This time around, however, we
think the sell-off was not because of the unwinding of
positions by global funds but the selling by Japanese active
pension funds.
While it is difficul t to pinpoint the reasons for selling by
Japanese active pension funds, we think investors need to be
aware of the possibility of a similar thing happening again. One
reason for this view is that the dissolution of employee
pension funds (EPFs) is about to take off on a large scale. Whether
or not what happened on 24–25 March was linked to
the return to the state of the substitutional portions of EPFs, the
risk remains that a similar thing could happen again. A
positive way of looking at what happened on 24–25 March would be to
say that it has served as 'disaster training' in
preparation for the risk of future unwinding of pension fund
positions.
Based on this view, we would like to stress the importance of
avoiding stocks that are the object of herding by active
funds. For example, if we assume that there is a
substantial unwinding risk for stocks in which only domestic
pension
funds are substantially overweight, then avoiding these stocks is
likely to be an effective strategy. This kind of strategy
would have performed well from mid-January 2014, and would have
generated positive returns without suffering a
negative impact on 24 March 2014.
Aki hi ro Murakami
3
Foreword We usually like to take a longer-term view in these notes,
but occasionally events take
over. On Monday 14 April, Momentum suffered a 2.6% intra-day fall
in Europe, a 2.6
standard deviation move. 1
Several aspects of this move are remarkable: both sector-neutral
and non-sector-neutral
Momentum suffered falls of a similar size. It has occurred without
a clear index-level
impact (the market has been both up and down over the course of the
three-week move,
and even over the course of the day on 14 April). It also has
happened without an
obvious ‘fundamental’ catalyst.
Conversations with clients have been focused on this issue for
several weeks now. The
Momentum sell-off is causing particular pain in levered
market-neutral strategies, and
this seems to be especially the case in those with shorter holding
periods where
momentum is used particularly often. The impact on long-only
investors has been more
muted.
Although a move of the size of 14 April probably has to be
caused by fund de-levering,
the ultimate cause is hard to pin down and, we suspect, may be the
result of several
forces. What we do know is that it leaves Momentum more cheaply
valued than it was
before. We upgraded our view on Momentum globally on 7 January
2014. As a result,
we have to be humble here as we have lost money for three weeks.
However, with
Momentum now trading cheaper than it ever has done before in 25
years in Europe, we
think this presents an opportunity. This is not 2009 or 2000,
periods when Momentum
sold off in a sustained way because on both those occasions the
factor was (very)
expensive. As a result, we double our position in Momentum. The
first section of this
note explores this recent fall and explains the upgrade.
At the same time as the Momentum sell-off in the US and
Europe there has been a large
factor move in Japan that has also caused severe dislocation to
some funds – again both
quant and non-quant. Another section of this note discusses this
move in detail. We think
that ultimately it is unrelated to the move in the other regions,
and is probably led by the
unwinding of domestic pension funds, but that means that such a
move could potentially
occur again.
This month we also have a piece that takes a longer term view of
value in a cross-asset
sense by applying it to rates as well as equities at the index
level. We show how a long-
short value strategy within rates outperforms a long-only
benchmark; we also show a
long-only version of the strategy that also outperforms on a
duration-matched basis.
Inigo Fraser Jenkins
4
Doubling up on momentum; this is not 2009 (or 2000) • The sharp
style moves seen in Europe, the US and globally since 21 March have
seen
Momentum sell off by 7%, 5% and 6% respectively to the close of
Monday 14 April.
These are very large moves. With Momentum having now given up all
of its gains since
we went long the style in early January, our call to be long
Momentum in Europe and
globally is losing money.
• As a result of the recent moves, the style is now cheaper than it
has ever been before.
In Europe, the high-momentum stocks are now trading at a 27%
discount to the low-
momentum stocks – this is unprecedented. Globally, the ‘winner’
stocks are also
trading at historical lows. Previous sustained Momentum sell-offs
have always
happened from much higher multiples.
• We consider several explanations that we have heard proposed for
the recent moves,
including the Nasdaq sell-off, various macro triggers, an EM-driven
rotation and fund
delevering. Although we think these have some merit, none seems to
offer a compelling
or complete explanation, nor a reason for these style trends to
continue.
• We see recent underperformance of Momentum as a buying
opportunity both in Europe
and globally. We are doubling up our position in Momentum in our
Global
Recommended Quant Portfolio.
The mysterious rotation
The factor moves we have seen over the past three weeks in Europe
and the US have
been sharp and, to most people we have spoken with, very
unexpected. The rotation – in
particular, the sell-off of Momentum in both regions and the rally
in Value, which was
particularly strong in the US – was especially surprising because
it seemed to have no
obvious catalyst, and initially unfolded against the background of
broadly flat markets
and reasonably unexciting (in either direction) corporate and macro
newsflow.
Fig. 1: US Momentum and Value
Source: Nomura Quantitative research 2
Fig. 2: European Momentum and Value
Source: Nomura Quantitative research
The moves started in early March, but intensified towards the end
of the month, with
Momentum falling by 100bp and 190bp, respectively, in Europe and
the US in the last
week of March alone. Strikingly, the ‘winners’ were also sold off
within sectors, ie, on a
sector-neutral as well as a market-wide basis (Figs. 1 - 3). Value
performed well at the
same time, although it outperformed significantly more in the US
than it did in Europe,
delivering a 2.9% return compared with Europe’s more modest
60bp.
2 . Long-short returns of factors by region. Value is defined as an
equally weighted
combination of price/book, trailing dividend yield and 12-month
forward P/E. Momentum
is defined as an equally weighted combination of a simple 12-month
price momentum
and the percentage change over the past two quarters in average
consensus 12-month
forward earnings estimates. The baskets are rebalanced
quarterly.
96
97
98
99
100
101
102
103
104
105
31-Dec-13 21-Jan-14 11-Feb-14 04-Mar-14 25-Mar-14
31 Dec 13 = 100 Co mp os it e Mo me nt um S N Co mp os it e Mo me
nt um
Composite Value SN Composite Value
96
97
98
99
100
101
102
103
104
105
106
31 Dec 13 = 100
94
96
98
100
102
104
106
108
110
31 Dec 13 = 100
Alla Harmsworth – NIplc
Fig. 4: European Factor perfo rmance
Source: Nomura Quantitative research. Risk is defined as an equally
weighted combination of standard deviation of returns, earnings
expectations and beta. Growth is defined as an equally weighted
combination of Internal Growth, five-year consensus expected growth
rate and three year forward growth rate. Quality is defined as an
equally weighted combination of ROE, credit rating, tax/pre-tax
income, EBITDA margin and change in the number of shares. The
baskets are rebalanced quarterly.
Fig. 5: Global Factor performance
Source: Nomura Quantitative research
The broad indices, confusingly, were flat at the same time, and
there was also a lack of a
clear pattern in terms of the performance of other factors or
sectors that would have
helped interpret the moves in Value and Momentum. The Value rally
would normally
signify a pro-cyclical bounce – yet in the US, where Value rallied
the most, Risk
underperformed slightly at the same time, while Quality
outperformed. At the sector level,
the better performers at the time were a mixture of Financials,
some non-commodity
cyclicals, Energy and classic defensives, such as Telecoms and
Staples. In Europe,
performance of other factors and of sectors was similarly
mixed.
After the initial bout of sharp declines, Momentum had
something of a rebound the
following week, only to sell off again, and more sharply, the week
after. Monday April 14
saw further large moves with Momentum falling by 3% and 1%
intra-day in Europe and
globally. One difference between the two episodes of
underperformance was that the
second time around, markets were down too, both in Europe and the
US. At the sector
level, things also felt more obviously defensive than in late
March. Staples and Utilities
rallied along with Energy, while most cyclicals did badly.
Importantly, Financials also
underperformed last week, in contrast with what happened in March.
The broader
decline in sentiment in the past week was reflected in our
Composite Sentiment
Indicator, which has just moved to levels indicating weak
sentiment, having fallen to the
lowest it has been since November 2012. That said, on Monday 14
April – the day that
saw some of the sharpest intra-day moves since the rotation started
three weeks ago –
markets were broadly flat overall.
Unsurprisingly, a number of fund managers have told us that they
suffered losses in
recent weeks. The underperformance has been most painful for some
levered market-
neutral funds – both fundamental and quant. Even for those who did
not lose money,
these recent moves have been a major focus and a source of concern,
and
understandably so. A Momentum sell-off is always a worrying event
because (and this is
one reason why so many investors are so uncomfortable with it) when
Momentum
investing goes wrong, it can go very wrong. An (admittedly extreme)
example is 2009,
when the strategy lost all the gains it had made in the previous
decade within months.
We explain below, however, that we think there is a crucial
difference between the factor
now and in the periods preceding any meaningful draw-downs in the
factor in the past 25
years – that difference is valuations. With Momentum currently
trading at literally
unprecedented lows, we do not expect a sustained sell-off in
high-momentum names –
historically, these have always happened when these names were very
significantly
more expensive than they are now.
A number of explanations for the recent factor moves have
been put forward. In the rest
of this note we discuss the ones we have heard most often. We think
that, although
some of them may have some merit, none offer a convincing enough
explanation for the
moves, nor do they give a reason to expect these style trends to
continue. We see
recent underperformance of Momentum as a buying opportunity on a
three-month view
and beyond, and are doubling up our position, both in Europe and
globally.
94
96
98
100
102
104
106
108
110
Composite Risk Size
31-Dec -13 14-Jan-14 28-J an-14 11-Feb-14 25-Feb-14 11-Mar-14
25-Mar-14 8-Apr-14
31 Dec 2013 = 100 Composite Growth Composite Quality
Composite Risk Size
A Macro catalyst?
The first obvious place to look for an explanation of a sharp
factor rotation like this is a
macro-related catalyst. It is possible that some of the move in
early March was indeed
caused by changes in macroeconomic sentiment. Figs. 6 and 7 show
that the initial sell-
off in Momentum and the outperformance of Value in the US in March
was accompanied
by a rising Economic Surprise Indicator, which has recently tended
to be positively
correlated with Value and negatively with Momentum. In Europe,
there has been
something of a positive relationship between macro sentiment – at
least on this measure
– and the performance of Momentum, and the late-March
Momentum underperformance
did indeed coincide with a decline in the surprise index there (Fig
8). (The relationship of
the indicator with Value in Europe is less clear (Fig. 9).)
Fig. 6: US Macro and Momentum perfo rmance
Source: Nomura Quantitative research
Source: Nomura Quantitative research
Source: Nomura Quantitative research
Source: Nomura Quantitative research
It is conceivable that in the US, the improving macroeconomic
sentiment through March
was at least partially responsible for the boost to value stocks as
they are more pro-
cyclical (through their greater exposure to Financials) while
hurting Momentum through
its greater exposure to more ‘stable growth’ sectors, which tend to
do best when the
economy peaks or starts to slow. In Europe, if Momentum suffered
from a decline in
macroeconomic sentiment that is suggested by the Economic Surprise
series, it would
have been through its ‘short’ side, ie, its large underweight in
the defensive sectors,
which did well at the time (see Fig. 10 for the sector composition
of Momentum in the
US and Europe). However, as we mentioned above, other sector moves
were far too
mixed for us to believe that these style moves were entirely
macro-driven. Besides, this
measure of macro sentiment does not help explain the second phase
of the Momentum
sell-off, with the surprise indicators in both regions either
moving in the ‘wrong’ direction
or by insufficient amounts at the time.
More generally, recent macroeconomic data and newsflow seem to have
been broadly
benign or neutral in both regions. The ECB indicated on Thursday 3
April that the
possibility of QE was unlikely to have a large impact at the point
of the announcement,
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.556
57
58
59
60
61
62
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
104
106
108
110
112
114
116
118
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
93
94
95
96
97
98
99
100
101
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
174
176
178
180
182
184
186
188
190
192
194
7
given that QE would be unlikely to happen unless and until European
macro data get a
lot worse. In the US, some positive labour market data have been
indicating that the
impact of bad weather on the economy is largely over – but again
this hardly seems
sufficient to have triggered the sudden and large-scale moves that
have taken place.
One specific macro ‘event’ that we have often heard mentioned in
connection with the
initial moves of late March is Janet Yellen’s speech on 18 March.
Again, we do not think
this holds a clue to what unfolded. The Value rally and the
Momentum sell-off did not
gather pace until about three days after the ‘hawkish’ speech was
delivered. Further, her
second, ‘doveish’ correcting speech did not prevent the second
phase of
underperformance – so the first one seems to have been an unlikely
trigger to begin
with.
In conclusion, there seems to have been no obvious macro-related
cause for the style
moves of the past three weeks. This is not to say that the moves
were entirely unrelated
to the macroeconomic environment – only that no particular aspect
of it seems to offer
an explanation for their sharpness and suddenness, or gives a
reason to expect these
moves to persist.
The US ‘new economy’ profit taking
The sell-off in the US Tech and Biotech stocks – which have
corrected sharply in the
past few weeks – is another commonly cited reason for the Momentum
correction that
we have been hearing. Again, this may have been a contributing
factor – both directly, in
so far as the Momentum baskets are exposed to the Tech/Biotech
names that have sold
off, and indirectly through potentially causing a ripple of
profit-taking behaviour in other
sectors and stocks that make up the style.
Momentum has a positive weight in Technology both in Europe and the
US – although
currently considerably more so in the US (Fig. 10). Furthermore,
within Tech, the style is
net long Internet stocks (with 1.5% exposure in the US and 1% in
Europe), which had
seen a substantial run in performance prior to March (Figs. 11 and
12). Momentum is
also exposed to Biotech, with a net 3.8% weight in the subsector in
the US and a 3%
weight in Europe. This is hardly surprising considering the
dizzying performance the
sector has delivered over the past couple of years.
Both Biotech and Internet stocks have declined sharply in the US in
recent weeks, down
10% and 8%, respectively, during March; both also fell further in
the first week of April.
The timing of the brunt of the sell-off in March did loosely
(though not precisely) coincide
with the timing of the Momentum sell-off.
Fig. 10: Composite Momentum – net sector exposure, % (long -
short)
Source: Nomura Quantitative research
Europe US
Source: Nomura Quantitative research
Source: Nomura Quantitative research
Source: Nomura Quantitative research
However, in Europe, the sell-off in the (tiny in size) Biotech
sector was much more
modest in magnitude, with the sector falling by 3% during March
compared with 10% in
the US; furthermore, the two or three stocks in Europe in the
Internet subsector actually
rose during March, with some of the bigger gains occurring just as
Momentum was
suffering towards the end of the month. So this particular
explanation does not seem to
offer much insight into what has gone on in Europe.
Even in the US, the overall exposure of the Momentum basket to
Biotech and parts of
Tech that may still be vulnerable to further correction seems far
too small for the
performance of the overall style to be dominated by these
stocks.
The Tech/Biotech unwinding may have also been having an indirect
effect on Momentum
by encouraging a bout of profit-taking to take place in other
sectors, both in the US and
in Europe. Almost tautologically, given that we are talking about a
Momentum correction,
there has been something of a tendency to sell whatever has gone up
the most across
sectors, as shown in Fig. 14, which plots the European sector
relative performance since
21 March, when the moves started, against their performance since
the beginning of last
year. Wider profit-taking behaviour is also evidenced by the fact
that, as we mentioned
above, the style has corrected on a sector-neutral as well as a
market-wide basis – that
is, ‘winners’ have also been sold off within each
broad sector.
50
100
150
200
250
300
350
400
Dec 08 = 100
0
50
100
150
200
250
0
100
200
300
400
500
600
700
IndexIndex
2
3
4
5
6
7
8
9
%
9
Fig. 14: European sector relative perfo rmance since Dec 2012 and
21 March 2014, %
Source: Nomura Quantitative research
Will this continue to affect the style? For one thing, the
profit-taking behaviour is likely to
have been exacerbated by concern regarding the forthcoming earnings
season, which in
the US has been precipitated by fairly gloomy company guidance and
a sizeable
downward adjustment in earnings expectations for US stocks in
recent weeks. We think
that, as the earning season gets underway, this effect will
diminish in importance, with
any potential negative surprises for the high-performing stocks
largely priced in. In
Europe, investors have also been increasingly conscious of the need
to see hard
evidence of positive earnings growth emerging in order for equities
to continue to
perform – so there is much at stake when 1Q reporting starts, and
this may have
contributed to the more defensive sentiment and profit-taking in
the market. Despite
some concerns regarding the impact of the currency on earnings,
however, most
investors, like us, seem to expect 1Q reports in Europe to be
positive.
There is much more to Momentum globally than the expensive Internet
and Biotech
stocks that are being sold off – as evidenced not just by the
straightforward read of the
sector composition of the baskets, but also by the fact that the
overall style on both sides
of the Atlantic is now exceptionally cheap as we show below. We
think that this should
support the performance of the factor, even if the unwinding of
some expensive parts of
the market were to persist.
-30% -20% -10% 0% 10% 20% 30%
MEDIA
FINANCIALS
TECHNOLOGY
UTILITIES
Emerging markets
The past two weeks have seen investors channel funds into emerging
markets (EM) for
the first time in six months. This is a significant event given how
extreme the negative
sentiment towards EM became in the past few months (indeed this
extreme pessimism
regarding EM, which manifested itself in a rare ‘double redemption’
of funds both from
the dedicated GEM funds and the total of individual EM country
funds, prompted us to go
long EM earlier this year (see Taking a long EM position, 4
February 2014). A number of
our clients have wondered whether the recent style rotation may
have somehow been
connected with this change in sentiment towards EM.
Fig. 15: Flows into GEM funds, USD m
Source: Nomura Quantitative research
If there is a link between the improving sentiment towards EM
assets and the relative
performance of styles in the US and Europe, we would expect this
relationship to play
out via the domestically listed (ie, US and European) stocks with
exposure to EM. In
Figs. 16-19 we look at the recent relationship between the daily
relative returns of the
EM-exposed basket of stocks and the performance of Value and
Momentum, in Europe
and the US.
European EM-exposed stocks indeed seem to have done well since
mid-March,
outperforming the market by over 200bp, while Value did well and
Momentum fell. This
relationship is in line with recent history – over the past year or
so, the relative
performance of the EM-exposed stocks has been positively correlated
with the
performance of Value in Europe, and negatively (if weakly) – with
the performance of
Momentum.
Fig. 16: Relative performance of European stocks with EM exposure
and the performance of Momentum
Source: Nomura Quantitative research
Fig. 17: Relative performance of European stocks wi th EM exposure
and the performance of Value
Source: Nomura Quantitative research
USD million
May 13 = 100May 13 = 100
Europe Momentum
93
94
95
96
97
98
99
100
85
90
95
100
105
110
115
120
May 13 = 100May 13 = 100
Europe Value
11
What about the US? As we would expect, with US companies in
aggregate much less
exposed to EM than companies in Europe, these stocks do not appear
to have moved
much in response to the resumption of flows into EM – and the
relationship between their
relative performance and style performance has been weak or
non-existent, both in the
past year and over the longer term. In particular, there has been a
zero correlation
between the relative performance of the EM-exposed basket in the US
and the
performance of Value over the past year or so, and a positive –
though not very strong –
relationship between the EM-exposed basket and Momentum. Certainly,
there was not a
sufficient move in these stocks in the past two or three weeks that
could explain the large
drop in Momentum that we saw.
Fig. 18: Relative performance of US stocks with EM exposure and the
performance of Momentum
Source: Nomura Quantitative research
Fig. 19: Relative performance of US stocks with EM exposure and the
performance of Value
Source: Nomura Quantitative research
What do we conclude from all this? In Europe, the EM-exposed stocks
have
underperformed over the past year – so in so far as some of these
stocks have become
very cheap and ended up populating the Value basket, their
outperformance when the
region started to benefit from inflows benefitted the performance
of Value. Conversely,
Momentum may have been short some of these names, so their
outperformance may
have hurt it. But with style constituents constantly changing, the
relationship between
EM-exposed assets and style performance has been neither strong nor
stable enough
historically for us to expect this effect to persist – even though
we do believe that the
positive flows into the region may be sustained. Moreover, the EM
effect does not shed
any light on what has gone on in the US – which, again, calls into
question how
important it could have been in Europe given the synchronicity of
the style moves in the
two regions.
Japan: a separate case
Japan has also seen some sizeable style moves over the past couple
of weeks, but both
their direction and drivers have, as is usual for this market, been
very different and
largely unrelated to what went on in the US and Europe.
Specifically, there was a
concentrated unwinding of positions by active funds on 24 March,
covered in detail in the
recent work by our Japan Quant team (see Concentrated position
unwinding on 24
March 2014, 28 March 2014), which thinks that the selling was
coming from the
Japanese active pension funds rather than global funds. In the two
days that followed,
there was strong underperformance of parts of Value (low P/E stocks
in particular sold
off strongly, in contrast to what we were seeing in the US and
Europe around the same
time) and Growth and Momentum also underperformed. Interestingly,
Momentum went
on to rebound in the days that followed to a greater extent than
Value – again in contrast
to what happened in the US and Europe.
98
100
102
104
106
108
110
May 13 = 100
98
100
102
104
106
108
110
112
May 13 = 100
Source: Nomura Quantitative research
Our Japan quant team argues that, although it is difficult to
pinpoint the reasons for the
concentrated selling on those days, investors need to be aware of
the possibility of the
same thing happening again. The dissolution of employee pension
funds (EPFs) is about
to take place on a large scale – and, whether or not what occurred
at the end of March is
linked directly to this process starting, the risk remains that a
similar thing could happen
again. Domestic pension fund portfolios tend to be tilted toward
Value factors, so the
longer-term effect of this is likely to be favourable for Momentum,
Growth and
Profitability-related measures and negative for the efficacy of
Value in Japan – a factor
that, historically, has performed the best there. Although the
process is likely to be
gradual, it would mean profound changes for quant investing in
Japan.
Doubling up on Momentum
None of the explanations that have been put forward to explain
recent style moves give
us sufficient reasons to expect further underperformance from here.
In fact, the sell-off
motivates us to double the size of our position, both in Europe and
globally. The factor
was already trading at historical lows, and valuations have now
become much more
extreme – after the recent rebalancing of the Momentum constituents
and taking recent
moves into account, Momentum in Europe has become even cheaper,
with the high-
momentum stocks trading at a 27% discount to low-momentum stocks on
a price/book
basis. For ‘winners’ to trade at a discount to the ‘losers’ is a
highly unusual event by the
very construction of the style, and has happened on only another
two occasions in the
past 25 years. A discount of this size, is, simply put,
unprecedented.
Globally, the factor is also very cheap, with the high-momentum
stocks trading at a
premium to low-momentum stocks, which is very low by historical
standards. Figs. 21
and 22 show the valuation of Momentum historically, with the ‘dots’
denoting the latest
valuations based on the new 2Q constituents and taking into account
the factor’s
performance to the close of Monday 14 April.
90
92
94
96
98
100
102
104
31 Dec 13 =100
Momentum (Non Sector Neutral)
Fig. 21: European Composite Momentum, price to book
Source: Nomura Quantitative research. Figure shows price/book of
the top quartile of stocks on our Composite Momentum measure
relative to the bottom quartile of stocks.
Fig. 22: Global Composite Momentum, price to book
Source: Nomura Quantitative research. Figure shows price/book of
the top quartile of stocks on our Composite Momentum measure
relative to the bottom quartile of stocks.
A look at the previous draw-downs that the factor suffered in
the past 25 years – which
we defined as periods of sustained underperformance amounting to
10% or more –
shows that on average these have happened when Momentum was nearly
four times
more expensive than it is today, and have never happened from
valuations at levels as
low as they are now. In fact, we showed in previous research that
historically, similar
valuation levels were always followed by the factor outperforming
on a six- and 12-month
view. (See Time to buy the winners, 7 January 2014.)
The factor should also benefit from continued support from the
macroeconomic
environment; with the global economy continuing to progress through
expansion, this
should be a favourable environment for the style. Lastly, its
composition remains pro-
cyclical, and in line with our fundamental sector views, which also
are unchanged, with a
preference for Financials and selected non-commodity cyclicals over
defensives such as
Consumer Staples and Utilities.
One key thing to stress is that for long-short tactical investors
with shorter holding
periods some caution needs to be exercised given that they seem to
have suffered
particularly badly from the recent moves and that the ultimate
cause of the sharp moves
like the one of 14 April is hard to pin down. Also, the move
appears to have been
amplified by fund delivering, which is hard to predict and could
potentially continue in the
very short term. As a call that is in large part driven by
valuations, this is for investment
horizons of three months and beyond.
Our other factor calls are unchanged. We continue to prefer Risk
over Quality, given that
quality stocks continue to trade at a large premium to risk stocks
while seeing
intensifying downgrades. We have recently downgraded Value to
neutral because of the
low valuation dispersion between value stocks and the rest of the
market as well as the
overly bullish analyst expectation for these stocks. Lastly, we
continue to discriminate
between growth measures (preferring the more cyclical expected
growth factors to
sustainable growth) and are neutral on Size.
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
%
%
14
Of bonds and bubbles – value outperforms in rates, as in equit ies
Value investing is nothing new. Especially in equities, value
investing is among the
oldest and most well-known strategies. As early as 1934, Benjamin
Graham and David
Dodd spelt out the principles of value investing in their text
‘Security Analysis’.
Academics such as Eugene Fama and Kenneth French have also
studied value formally.
For example, Fama & French (1992) used value as part of their
three-factor model of
stock returns. Legendary investors such as Warren Buffett and Seth
Klarman famously
use the principles of value to guide their investment
decisions.
At its core, value investing is simple: buy assets that are
cheap, sell assets that are
expensive. Or as what Warren Buffett has observed in one of his
letters, “whether we’re
talking about socks or stocks, I like buying quality merchandise
when it is marked down”.
In this article, we focus on value in interest rates. We find that
using techniques
fundamentally similar to those used in equities can help rates
investors identify value,
across both curves and currencies. We discuss the following
points:
• Value exists in both equities and interest rates.
• Carry in interest rates is analogous to earnings yield (E/P) in
equities. Just as stocks
with high earnings yield outperform those with low earnings yield,
curve points with high
carry outperform those with low carry.
• Comparing earnings yield both relative to its own history
(time-series) and across
stocks (cross-sectional) is important to place current valuations
in context. Similarly,
comparing carry on both time-series and cross-sectional basis is
key.
• A long/short value strategy in rates outperforms a long-only
benchmark. Being free to
trade the best points of value across curves and currencies,
without constraining the
strategy to hold pre-specified pairs, can help make the strategy
robust.
• Bubbles develop gradually, be it in equities or rates. Equities
and rates don’t sell off
when they are cheap – crashes occur in the context of stretched
valuations. Value-
signals can help track a build-up in overvaluation and identify
when sell-offs in rates are
more likely.
• A long-only version of the value strategy outperforms a long-only
benchmark on a
duration-matched basis. By going long the cheapest curve points
across countries, the
strategy beats a static benchmark that holds the entire
market.
• A short-only version of the value strategy can help hedge rising
rates. It can also be
cheaper than being the short the entire market as it identifies the
most expensive points
to be short rather than shorting all points.
• Value can outperform when momentum suffers. Momentum tends to
underperform
when rates are close to zero and markets range-bound. In such
trendless conditions,
value or mean-reversion trades tend to outperform.
Value works in both equiti es and rates
Assets that offer good value outperform those that do not.
This appears to be equally
valid across equities and rates. In Fig. 23, we show performance of
S&P 500 and G4
interest rate markets conditional on valuation. Both equity and
rates outperform when
valuations are attractive relative to when valuations are poor. The
question then is: how
to measure ‘value’? We will first look at some common measures of
value in equities and
then extend the same fundamental approach to rates.
Tam Rajendran - NIplc
Source: Nomura research, Bloomberg
Value measures in equities
Common measures of value in equities compare the market price of a
stock with
fundamental measures such as book value, forward or trailing
earnings, and dividends, etc,
usually aggregated across a few quarters to smooth out short-term
shocks. These are
sometimes further adjusted for business cycle and inflation
effects, and compared with
long-term levels to gauge the valuation perception of current
investors.
One common measure is the cyclically-adjusted price-earnings ratio
(or CAPE; Campbell
and Shiller, 1988, 1998, 2001). Popularised by Nobel prize-winning
economist Robert
Shiller, CAPE (sometimes referred to as Shiller’s P/E) looks at the
ratio of real price to
real earnings smoothed over a 10-year period. A high P/E ratio
indicates that current
market prices are high relative to earnings, perhaps suggesting
overvaluation. Similarly,
a low P/E may suggest the market is undervalued. Fig. 24 shows
historical Shiller’s P/E
for the S&P composite since 1900. Shiller’s P/E has been a
robust indicator of valuation
in this period. The ratio has tended to be relatively high before
large sell-offs, like in
1929, 1966, 2000 and more recently, in 2007. These were times of
equity bubbles, when
prices were extremely high, and stocks expensive and overvalued. In
contrast, the ratio
has been relatively low, indicating cheapness, just before
significant bull markets, as in
1932, 1982 and 2009.
0.0%
0.1%
0.2%
0.3%
0.4%
0.5%
0.6%
0.7%
0.8%
0.9%
Equity
A v e
r a g
e m o
n t h
l y e x
c e s
s r e
t u r
n s
( % ) Good
Value
0
1
2
3
4
5
6
7
Rates
A v e
r a g
e m o
n t h
l y e x
c e s
s r e
t u r
n
( b p
)
Poor Value
40
400
0
10
20
30
40
50
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010
S & P
c o m p
o s
i t e
i n d e
x
S h
i l l e
r ' s
P / E
1929
1966
2000
2007
16
A related and widely-used valuation measure is the earnings
yield. It is calculated as the
ratio of earnings to price, ie, the inverse of P/E. Fig. 25 shows
historical earnings yield for
the S&P composite since 1900. Stating valuation in yield terms
offers a quick way of
comparing potential return both across individual stocks and
different asset classes 3 .
Indeed, typical equity value strategies compare earnings yield both
in a time-series and
cross-sectional sense.
Fig. 25: Low earnings yields means expensive equities
Source: Nomura research, Shiller’s website:
www.econ.yale.edu/~shiller/
The relationship of earnings yield to expected returns has been
documented in Basu (87)
and Ball (92) among others. Historically, earnings yield has proven
to be a good proxy
for expected returns. High earnings yields tend to precede higher
expected returns and
vice-versa. In Fig. 26, we show one-year ahead S&P returns
conditional on current level
of earnings yield.
Fig. 26: Equities ou tperform when earnings yi elds are high
Source: Nomura research, Shiller’s website:
www.econ.yale.edu/~shiller/ . Analysis based on data from 1900 to
2014.
We find that one-year ahead stock returns are the highest when
earnings yields are in
the top quintile and weakest when earnings yields are in the bottom
quintile. In other
words, measuring earnings yield relative to its own historical
average can help place
current valuations in context. Earnings yields above historical
average may mean
3 Dividend yield is a more direct measure, but is inadequate in
practice owing to tax
considerations and variations in corporate payout policy. For
example, companies may
favour stock buy-backs over dividends for issues unrelated to
intrinsic valuation.
40
400
0.00
0.05
0.10
0.15
0.20
0.25
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010
S & P
c o m p
o s
i t e
i n d e
x
E a r n
i e
l d
1929
1966
2000
2007
0
4
8
12
16
S u
t S &
P r e a
l r e
t
t i m
e
t + 1
( %
, p . a
. )
1y ahead
markets are undervalued (and vice-versa). Before using earnings
yields to compare
different stocks or different markets, it may be important to
standardise the indicator to
adjust for idiosyncrasies. Adjusting for the long-term average
makes cross-sectional
comparisons consistent and robust.
In Fig. 27, we show a global equity value strategy that goes long
stocks with high
earnings yield (cheap) and short those with low earnings yield
(expensive). This strategy
has outperformed the MSCI World Equity index over the long
sample.
Fig. 27: Global equity value strategy
Source: Nomura research, Bloomberg
Value measure in rates: carry as ‘earnings yield’
Carry in interest rates is analogous to earnings yield in stocks.
Carry measures the
return on a bond assuming rates do not move. The relationship
between current level of
carry and future bond returns is similar to that of earnings yield
and stocks: points on the
curve with high carry tend to outperform those with low
carry.
Similar to equity measures, adjusting carry for the long-term
average can be important.
Carry measured relative to its own historical average (‘relative
carry’) can help identify
times of potential overvaluation and undervaluation in interest
rates and place current
levels in context. When carry is relatively low, it suggests future
returns for bonds are
likely to be weak. Similarly, when carry is relatively high, future
bond returns are likely to
be strong. In other words, low relative carry implies rates are
expensive and offer poor
value, while high relative carry implies rates are cheap and offer
good value. Fig. 28
presents the evidence across G4 markets. Rates have outperformed
during period of
good value across curves and currencies.
Fig. 28: Rates outperform in periods of good value
Source: Nomura research, Bloomberg
Global Equity Value
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
5y 10y 20y 30y 5y 10y 20y 30y 5y 10y 20y 30y 5y 10y 20y 30y
A v g .
t u r n
Top third Bottom third
18
We can now construct a value strategy in rates along the lines of
equity value strategies.
We trade 5y, 10y, 20y and 30y interest rate swaps across the USD,
GBP, EUR and JPY
curves. We measure carry as the 1m rolldown relative to its
longer-term average at each
curve point. We then compare this indicator across the 16 possible
curve points and go
long the top two points and short the bottom two points.
Fig. 29: Looking for value: searching across curves and
currencies
Source: Nomura research. Positions shown current as of 14th April
2014
Such a value strategy has outperformed a long-only G4 government
bond index
historically. The rates value strategy has higher Sharpe and Calmar
ratios over the long
sample and in recent periods. It also has favourable skew
characteristics.
Fig. 30: Rates value outperforms long-only – higher Sharpe and
Calmar ratios, better skew
Source: Nomura research. Performance statistics are based on
monthly returns from Jan 1992 to Feb 2014
We also scale the long/short exposure to a fixed-duration target to
ensure the strategy
remains duration neutral. As a result, the final strategy has very
little overall duration
exposure. The scatter plots below show monthly excess returns vs
monthly changes in
10y yields across G4 markets. The long-only government bond index
shows typical long-
duration behaviour – rising yields, negative returns and
vice-versa. The rates value
strategy does not have any significant bias.
Low value High value
USD EUR GBP JPY
5y Long Short Long
1992 1997 2002 2007 2012
E x c e
s s r e
t u r n
i n d e
x
Rates value strategy
0.0
0.4
0.8
1.2
1.6
Entire sample Last 10 years Last 5 years Last 3 years
S h a
r p e r a
t i o
Rates value strategy
0.0
0.4
0.8
1.2
1.6
2.0
Entire sample Last 10 years Last 5 years Last 3 years
C a l m
a r r a
t i o
Rates value strategy
-1.0
-0.6
-0.2
0.2
0.6
1.0
Entire sample Last 10 years Last 5 years Last 3 years
S k e
w
19
Fig. 31: Better returns, no bias – rates value outperforms despite
flat-duration exposure
Source: Source: Nomura research, Bloomberg. Analysis based on data
from 1992-2014
This can also be seen from the effective duration of the two
strategies, calculated as the
negative coefficient of returns regressed on changes in yield. As
expected, the long-only
bond index has a significantly positive duration of close to 5,
while the value strategy is
practically flat duration exposure.
Searching for value across cu rves and currencies
In equities, earnings yield is a powerful tool used to estimate
relative value and identify
opportunities for outperforming the benchmark. Valuation based on
CAPE across
countries may be more robust as a metric for future returns
compared with single-country
implementations (Faber 2012). By diversifying across a larger set,
one is more likely to
diversify away idiosyncratic effect.
Similarly in rates, searching for value across curves and
currencies provides the extra
diversification for long-term consistent outperformance. Allowing
the strategy to choose
the best and worst curve points without constraining it to choose
among pre-specified
cross-country pairs lets it adapt to changing yield curve dynamics.
Also, comparing
current levels of carry relative to history helps place valuation
in context and avoids
biased positioning. Using absolute levels of carry to position can
prove to be misleading
and lead to an undiversified strategy. Fig. 32 shows the percentage
of time a particular
curve point was selected. We find that the strategy that looks at
relative carry is in
general well diversified. A strategy that uses absolute levels of
carry would have never
picked most of the points and would have been heavily
undiversified.
-3
-1.5
0
1.5
3
-1 -0.5 0 0.5 1 M o
n t h
l y e x
c e s
s
r e
t u
r n s
( % )
-3
-1.5
0
1.5
3
-1 -0.5 0 0.5 1 M o
n t h
l y e x
c e s
s
r e
t u
r n s
( % )
Effective duration
20
Fig. 32: Comparing carry against its own histo ry provides context
to valuation, helps diversification
Source: Nomura research, Bloomberg. Analysis based on daily data
from 2000, when data from all curve points become available.
Fig. 33: Using absolute levels of carry ignores current context,
leads to biased positioning
Source: Nomura research, Bloomberg. Analysis based on daily data
from 2000, when data for all curve points become available.
Evolving macro regimes can significantly affect the initial
assumptions of a strategy. One
such example is the definition of the ‘belly’ of the curve. In the
USD curve, it is common
to use the 5y point as the belly and reasonably so. This might not
be the case in Japan.
Starting with the introduction of the zero interest rate policy by
the Bank of Japan in
1999, the Japanese yield curve has become extremely flat out until
the 10y point. The
front-end of the JPY curve can then be said to extend until the 5y
or perhaps even
beyond until the 10y point. In such a scenario it might be
inaccurate to treat the
Japanese 5y point as the ‘belly’ and expect it to have dynamics
similar to the US 5y
point. With the Fed keeping short-term rates low, even US rates
might be headed in the
same direction as Japan. The USD ‘belly’ too might move down the
curve as front-end
rates become flatter.
Macroeconomic conditions change and curve behaviour reflects these
changes. A rates
strategy, which has fixed assumptions about curves in its design,
can underperform as
reality starts to differ from its initial assumptions. A strategy
that is dynamic and is free to
move across curves and currencies can be more adaptive to the
market.
Using value to identify sell-offs
Be it equities or rates, bubbles develop gradually; they rarely
happen out of the blue.
Crashes occur when assets are expensive and valuations stretched.
Equities do not
crash when they are cheap. Similarly, rates sell off not when they
are cheap, but when
they have been consistently expensive and overvalued. Often, in
periods of such
overvaluation, tell-tale signs show up in the data. Rules-based
value strategies that stick
to following the data can help track a build-up in valuation. It
can be difficult to time such
events perfectly, but sticking to simple and consistent valuation
methodologies, and
having checks that compare along both time-series and
cross-sectional basis can help
make the strategy robust.
We look at three cases of sell-offs in interest rate markets, all
of which were thought to
be ‘sudden’ or ‘unexpected’ at that time: the inflation driven
sell-off in the US in 1965, the
February 1994 sell-off in the US, and the so-called ‘VaR shock’ in
Japan in June 2003.
But applying the techniques we describe in the previous sections,
it seems rates were
indeed overvalued and were likely to crash in each of these
cases.
The charts below show the 10-year yields in the respective
countries. The colour of the
columns shows traded positions in that particular country; the
columns in the top half are
long positions and those in the bottom half are the short
positions.
Fig. 34: Case study 1: US sell-off i n 1965
Source: Nomura research, Bloomberg. Due to unavailability of swap
rates data, bond yields have been used as proxy.
At the start of 1965, inflation in the US (gold line) was at
a moderate 1% y-o-y. By the
end of the next year, it had reached around 3.5%. The value
strategy was long the 20y,
and then later, the 5y point (blue and grey columns in the top
half). But soon after, on the
basis of relative carry, USD rates began to appear expensive. In
the latter half of 1965,
the strategy shifted to being short USD, just as yields began to
rise. From then on, it was
consistently short USD rates through the 10y and 20y points (light
and dark blue
columns in the bottom half). Positioning based on value would have
helped capture most
of the sell-off.
Source: Nomura research, Bloomberg.
February 1994 saw the start of one of the largest sell-offs in US
rates. 10y bond yields
rose by over 200bp in the course of that year. Although the
sell-off was deemed to be
unexpected by most market participants, indicators of value told a
different story. From a
valuation perspective, the strategy found US rates to be most
expensive among the
sixteen points it looks at. Accordingly, the value strategy was
short the 5y and 30y points
(grey and purple columns in the lower half). It continued to be
short USD rates
throughout the year, switching to short 5y and 10y later and
benefited from the sell-off.
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
5.5
-1
0
1
USD 20y
USD 10y
USD 5y
USD 10y (%, rhs)
US CPI (yoy, %)
S h o
r t p o
s
i t i o
n s i n
t h e
U S
L o n g
p o s
i t i o
n s i n
t h e
U S
5.5
6.0
6.5
7.0
7.5
8.0
8.5
-2
-1
0
1
2
USD 30y
USD 20y
USD 10y
USD 5y
USD 10y yield, rhs, %
S h o
r t p o
s
i t i o
n s i n
U S D
Fig. 36: Case study 3: Japan sell -off in 2003
Source: Nomura research, Bloomberg
The year 2003 started benignly for Japanese rates. The 10y point
had continued its rally
and had fallen by around 35 bps in the first few months, reaching
its all-time low in June.
But the strategy found valuations to be stretched and JPY rates
expensive compared to
their own history and across markets. Based on this, it went short
the 20y and 30y points
(dark blue and purple columns in the lower half). In June 2003,
rates rose sharply and
sold-off by over 100 bps in a matter of months and reached a level
of 1.6% in September
2003. The strategy, which was positioned short throughout this
episode, benefited from
this sell-off.
The strategy continued to find JPY rates overvalued and remained
short as rates
continued to rise into 2004. The sell-off finally cooled around the
middle of 2004.
Interestingly, JPY rates began to look attractive at this point.
The strategy found them to
be relatively cheap and undervalued. Towards the end of 2004, just
as rates started to
rally, it went long JPY rates. As shown, the strategy went long the
20y and 30y points
predominantly (dark blue and purple in the top half) and the 10y at
other times (light
blue).
Using value to outperform a long-only benchmark
The strategy described in the previous sections is long/short,
based on value. It receives
interest rate swaps at the two cheapest curve points and pays
interest rate swaps at the
two most expensive curve points. Receiving interest rate swap is
equivalent to being
long a bond. But it is possible to only use the long part of the
strategy. In Fig. 37 we
show a long-only variation of the value strategy described before.
Based on relative
carry, the strategy searches for points across curves that offer
the best value or are
‘cheapest’. It then buys these points, matching the duration to
that of the benchmark.
Such a strategy has outperformed a long-only G4 government index.
It has improved
returns and lowered the drawdowns, resulting in higher Sharpe and
Calmar ratios.
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
-2
-1
0
1
2
JPY 30y
JPY 20y
JPY 10y
JPY 5y
i t i o
n s
i n
i t i o
n s
i n
Fig. 37: Long -only value strategy outperforms a long-only
benchmark
Source: Nomura research, Bloomberg. Both strategy and long-only
index have been scaled to have the same annualised volatility for
easier comparison.
The outperformance is largely attributable to two things the
strategy does differently.
First, it trades interest rate swaps instead of bonds. Trading
swaps offers an immediate
pickup in yield over trading bonds. Second, it dynamically looks
for the best points to be
positioned long across the curve. Selectively being long the part
of the curve that is
cheapest can help outperform a benchmark that statically holds the
entire curve. After
all, the benchmark does not distinguish between expensive and
cheap, between good
value and poor value.
Using value to hedge rising rates
The counterpart to the long-only part of the value strategy is to
trade only the short
positions. The short-only value strategy sells the two most
expensive curve points across
curves, providing a cheaper way to hedge rising rates than going
short the entire market.
In Fig. 38 we compare such a strategy to a short-only G4 government
bond index (the
negative of the long-only index used in the previous section). As
before, the strategy has
been targeted to have a duration similar to that of the bond
index.
90.0
Long-only value strategy (top 2) Long-only G4 govt. bond
index
1992-2014 2000-2014 2011-2014
Volatility (%) 4.55 3.17 5.11 3.38 4.49 3.64
Sharpe 0.95 0.85 0.93 0.73 1.19 0.93
MDD 8.55 8.42 7.22 6.84 4.28 4.60
Calmar 0.51 0.32 0.66 0.36 1.25 0.73
24
Fig. 38: Short-only value strategy out performs a short -only
index
Source: Nomura research, Bloomberg. Both strategy and long-only
index have been scaled to have the same annualised volatility for
easier comparison.
The short-only value strategy bleeds much less than the short-only
bond index. It also
has much lower drawdowns and better returns. Fig. 39 highlights the
source of gains.
When rates are rising, it delivers returns almost on par with being
short the entire market.
But, more importantly, when rates are falling, it loses only half
as much being short the
market. It offers a cheaper way to protect against rising
rates.
Fig. 39: Short-only value offers a cheaper way to hedge rising
rates
Source: Nomura research, Bloomberg. Analysis based on monthly
returns from 1992-2014.
Using value to outperform in trendless markets
Like a long straddle position, momentum requires large moves in
either direction to
perform. If interest rates remain low and markets range-bound, such
trading rules will
suffer. For example, before Bank of Japan introduced its zero rate
policy in 1999,
momentum strategies worked well in JGB futures. Since 1999,
however, such strategies
have failed to perform. Instead, value strategies that trade on
mean-reversion in JGB
futures have started to outperform. The figure below shows the
performance of
momentum and value strategies in JGB futures.
50
Short-only value strategies (bot tom 2)
Short-only G4 govt. bond index
1992-2014 2000-2014 2011-2014
Volatility (%) 3.40 3.17 3.21 3.38 2.70 3.64
Sharpe -0.24 -0.85 -0.23 -0.73 0.00 -0.93
MDD 23.38 46.74 16.34 31.67 7.50 13.81
Calmar -0.04 -0.06 -0.04 -0.08 0.00 -0.24
-1
-0.6
-0.2
0.2
0.6
1
Months of negative LO returns Months of positive LO returns
A v g .
m o n
t h
l y e x
c e s s
r e
t u r n
s
( % )
25
Fig. 40: Momentum has underperformed while Value has outperformed
in Japan post- ZIRP
Source: Nomura research, Bloomberg. Both strategies have been
scaled to have the same volatility for easier comparison
This has been evident in global markets more recently. With global
yields at historical
lows, interest rate markets have remained trendless. As a result,
trend-following
strategies like momentum have underperformed. In contrast, value
strategies have done
very well. The chart below shows performance of value and momentum
strategies
across G4 markets last year.
Fig. 41: Value outperformed Momentum in G4 rates in 2013
Source: Nomura research, Bloomberg
Conclusion
Value strategies in equities have been popular for decades. Typical
value strategies in
equities use earnings yield as a measure of valuation. Earnings
yield compared to its
own history and across markets helps place current valuations in
context and to identify
expensive/cheap stocks. In interest rates, carry is similar to
earnings yield. Just like in
equities, comparing carry to its own history and across curves can
help identify points of
overvaluation or undervaluation. A value strategy in interest rates
based on such a
measure has outperformed over time, delivering higher returns and
lower drawdowns
compared to a long-only benchmark.
Value strategies can be useful in many different ways. By tracking
measures of value the
strategy can help identify times when valuations are stretched and
rates likely to sell-off.
A long-only value strategy can also help beat a long-only
benchmark by selecting points
that are cheap and undervalued to go long instead of being long the
entire market. Its
opposite, the short-only value strategy can help hedge rising rates
better than a strategy
that is short the entire market, as it shorts only points that are
overvalued and expensive.
80
100
120
140
160
180
ZIRP begins
JGB momentum
JGB value
G4 rates value
G4 rates momentum
26
Also, when rates are close to zero and markets range-bound,
momentum strategies may
underperform. However, in these trendless environments, value
strategies tend to
outperform.
The essential truth of value – to buy assets that are cheap and
sell assets that are
expensive – is universal. It works in equities and also in rates.
Borrowing fundamental
techniques of valuation from equities and applying them to interest
rates can help design
27
References
• Asness, C.S., Moskowitz, T.J., and Pedersen, L.H. (2013), “Value
and Momentum
Everywhere,” in Journal of Finance 68, 929-985
• Ball, R. (1992), “The earnings-price anomaly,” in Journal of
Accounting and Economics
15, 319-345
• Basu, S. (1977), “Investment Performance of Common Stocks in
Relation to Their Price
to earnings Ratios: A Test of the Efficient Market Hypothesis,” in
Journal of Finance 32,
663-682
• Campbell, J.Y., and Shiller, R.J. (1988), “Stock Prices,
Earnings, and Expected
Dividends,” in Journal of Finance 43, 661-676
• Campbell, J.Y., and Shiller, R.J. (1998), “Valuation Ratios and
the Long-Run Stock
Market Outlook,” in Journal of Portfolio Management 24, 11-26
• Faber, M.T. (2012), “Global Value: Building Trading Models with
the 10 year CAPE,” in
Cambria Quantitative Research, No.5, August 2012
• Fama, E.F. and French, K.R. (1992), “The cross-section of
expected stock return,” in
Journal of Finance 47, 427-465
• Frazzini, A., Kabiller, D. and Pedersen, L.H. (2013), “Buffett’s
Alpha,” NBER Working
Paper No. w19681
28
Concentrated position unwinding on 24 March • From 24 March 2014
through the morning’s trading session on 25 March, actively
managed Japanese equity funds faced a relatively strong headwind,
especially in large
cap universes. Strategies based mainly on analyst earnings
forecast-related factors
and stocks with high active fund ownership ratios produced negative
returns of less
than -3% over this period. It is highly probable that, for some
reason, there was a
concentrated spate of selling for position unwinding purposes by
active funds on those
two days. The sell-off appears to have come to an end before the
start of the afternoon
trading session on 25 March, but we think there is still a need for
caution.
• We all remember the "quant headwind" in August 2007, when selling
pressure was
similarly concentrated on certain investment styles, such as
E/P-based factors, and the
performance of a number of factors plummeted at the same time in
multiple markets
around the world. This time around, however, we think the sell-off
was due not to the
unwinding of positions by global funds but to selling by Japanese
active pension funds.
• While it is difficult to pinpoint the reasons for selling by
Japanese active pension funds,
we think investors need to be aware of the possibility of a similar
thing happening
again. One reason for this view is that the dissolution of employee
pension funds
(EPFs) is about to take off on a large scale. Whether or not what
happened on 24–25
March was linked to the return to the state of the substitutional
portions of EPFs, the
risk remains that a similar thing could happen again. A positive
way of looking at what
happened on 24–25 March would be to say that it has served as
"disaster training" in
preparation for the risk of future unwinding of pension fund
positions.
• Based on this view, we would like to stress the importance of
avoiding stocks that are
the object of herding by active funds. For example, if we assume
that there is a
substantial unwinding risk for stocks in which only domestic
pension funds are
substantially overweight, then avoiding these stocks is likely to
be an effective strategy.
This kind of strategy would have performed well from mid-January
2014 onward, and
would have generated positive returns without suffering a negative
impact on 24 March
2014.
1. Unwinding of Japanese large caps
On 24 March and in the morning trading session on 25 March,
managers of active
Japanese equity funds were hit by a headwind. In this chapter we
look at the intraday
performance of the major quant factors for Japanese equities on 24
and 25 March. The
analysis set out below shows that large caps suffered the greatest
losses and strategies
based mainly on analyst earnings forecast-related factors and
stocks with high active
fund ownership ratios produced returns of around -2% on 24 March
and around -1% in
the morning session on 25 March. It is highly probable that, for
some reason, there was
a concentrated spate of selling for position unwinding purposes by
active funds on those
two days.
1.1 Headwind for active fund managers
We look first at the performance of key quant factors on 24 March
2014. We divided our
two universes—large caps (TOPIX 500 stocks) and small caps (listed
stocks with market
caps of ¥10–200bn)—into quintiles based on factor values as of the
beginning of March
2014. For each universe, we then calculated the daily return
(equally weighted basis) on
24 March on a strategy comprising long positions in the stocks in
the quintile with the
highest factor values and short positions in the stocks in the
quintile with the lowest
factor values.
Fig 42 shows the results. Estimated E/P, a valuation-related
factor, and expected ROE, a
growth-related factor, generated substantial negative returns,
particularly for large caps.
Estimated E/P generated returns of -1.96% and -1.25% for large caps
and small caps,
respectively, while expected ROE generated returns of -1.76% and
-0.8%, respectively.
The active fund ownership ratio, which is the supply and demand
factor that we think
requires the most attention, generated the lowest daily return for
large caps on 24 March
of all the factors we examined, at -2.4%.
Akihiro Murakami - NSC
29
What happened in the Japanese equity market on 24 March, just ahead
of the fiscal
year-end, has no doubt caused problems for fundamentals-based
active fund managers.
Not only did strategies based on analyst forecasts, which many
active fund managers
favor, fail to function, but active fund managers also saw a sharp
deterioration in the
performance of many of the stocks in their portfolios.
Fig. 42: Factor returns on 24 March 2014
Note: Based on TOPIX 500 stocks for large caps and listed stocks
with market caps of ¥10–200bn for small caps, as at the time of
rebalancing at the beginning of each month. For each factor, we
divided each universe into quintiles based on factor values , and
calculated the daily return (equally weighted basis) on 24 March,
including dividends, on a strategy comprising long positions in the
stocks in the quintile with the highest factor values and short
positions in the stocks in the quintile with the lowest factor
values. We did not take sector allocation into account. Top three
factor returns are shown in red, bottom three are shown in
blue.
Source: Nomura
1.2 Selling for position unwinding purposes ended before afternoon
session on
25 March
We look next at intraday performance on 24 and 25 March. Our
analysis indicates that
selling that appeared to be for position unwinding purposes by
active funds, which
started when the market opened on 24 March, came to an end before
the afternoon
session on 25 March.
We now look at intraday factor performance in order to analyze the
situation in more
detail. Our analysis methodology is the same as in Section
1.1.
Fig. 43 shows intraday factor returns on 24 March 2014. While the
TOPIX rose sharply
and the mood in the market as a whole appeared to be, if anything,
rather optimistic
(Fig.43 (c)), active fund managers were having a difficult
time.
Fig. 43 (a) shows how low-P/E stocks were sold off from the start
of trading on 24 March
and saw share price declines of around 1% in the morning session
and another 1% or so
in the afternoon session. Similarly, Fig. 43 (b) shows that stocks
with high active fund
ownership ratios suffered further price declines in the afternoon
session.
Factor return as of 24 March 2014 (%)
Factor Large cap (Ranking) Small cap (Ranking)
Valuation Estimated E/P Undervalued – overvalued -1.96 (27) -1.25
(28)
B/P Undervalued – overvalued 0.39 (5) 0.20 (15)
Estimated dividend yield High – low -0.69 (17) -0.68 (26)
Time series normalized E/P Undervalued – overvalued 0.70 (4) 0.38
(9)
Time series normalized B/P Undervalued – overvalued 0.77 (3) 0.67
(6)
Growth Expected ROE High – low -1.76 (26) -0.80 (27)
Expected pretax profit margin High – low 0.85 (2) 0.62 (7)
Recurring profit growth High – low -1.06 (20) 0.35 (10)
Quality Total accruals Quality low – high -0.02 (7) -0.01
(19)
Merton default probability High – low -0.13 (10) 0.77 (4)
Size Market cap Large – small -1.21 (23) 0.77 (3)
Historical return Historical 1-month return High – low -0.04 (8)
0.24 (13)
Historical 3-month return High – low -0.20 (11) -0.59 (24)
Historical 6-month return High – low -0.82 (19) 0.02 (17)
Historical 12-month return High – low -1.09 (21) 0.22 (14)
Risk Fundamental beta High – low -1.11 (22) 1.33 (1)
Specific risk High – low -0.44 (14) 0.14 (16)
60-day volatility High – low -0.57 (16) 1.29 (2)
Supply/demand Active fund ownership ratio High – low -2.40 (28)
-0.01 (20)
Nonresident ownership ratio High – low -1.38 (24) 0.75 (5)
Pension investment trust ownership ratio High – low -1.46 (25) 0.27
(12)
Individual investor ownership ratio High – low 1.02 (1) 0.34
(11)
Net equity finance High – low -0.09 (9) 0.01 (18)
Consensus forecast Change in recurring profits High – low 0.08 (6)
0.58 (8)
Change in rating High – low -0.45 (15) -0.63 (25)
Change in target price High – low -0.33 (12) -0.06 (21)
Earnings surprise Y-y change in quarterly progress High – low -0.77
(18) -0.53 (23)
30
This trend continued into the next day, with similar phenomena
observed at least through
the morning session of 25 March as on 24 March (Fig. 44 (a) and
(b)). The return on
estimated E/P fell a further 1% or so in the morning session, while
the active fund
ownership ratio factor generated a return of less than -1%.
However, these trends were
reversed immediately after the start of trading in the afternoon
session on 25 March, and
both low-P/E stocks and stocks with high active fund ownership
ratios were bought back.
The selling thought to be for the purpose of position unwinding by
active fund managers
that began at the start of trading on 24 March thus appears to have
come to an end
before the afternoon session on 25 March. However, we think
investors need to remain
aware of the risk that a similar sudden deterioration in the
supply/demand balance might
occur again.
Fig. 43: Intraday factor retu rns on 24 March 2014
Estimated E/P
c) TOPIX
Note: Based on a universe of TOPIX 500 stocks. For each factor, we
divided the universe into quintiles based on factor values at the
beginning of March 2014. We then calculated the intraday return
(equally weighted basis) and intraday turnover ratio on 24 March on
a strategy comprising long positions in the stocks in the quintile
with the highest factor values and short positions in the stocks in
the quintile with the lowest factor values. We did not take trading
costs into account. Analysis is based on historical data and does
not guarantee future performance.
Source: Nomura
x 10 -8 (%)
09:00 09:30 10:00 10:30 11:00 11:30 12:00 12:30 13:00 13:30 14:00
14:30
- 2
- 1.5
- 1
- 0.5
0
Intr aday factor return (lhs) and in traday turnover ratio
(rhs)
0
1
2
3
4
5
6
7
8
9
x 10 -8(%)
09:00 09:30 10:00 10:30 11:00 11:30 12:00 12:30 13:00 13:30 14:00
14:30
-2
-1.5
-1
-0.5
0
Intraday factor return (lhs) and in traday turnover ratio
(rhs)
09:00 09:30 10:00 10:30 11:00 11:30 12:00 12:30 13:00 13:30 14:00
14:30 0
0.5
1
1.5
2
Fig. 44: Intraday factor retu rns on 25 March 2014
(a) Estimated E/P
(c) TOPIX
Note: Based on a universe of TOPIX 500 stocks. For each factor, we
divided the universe into quintiles based on factor values at the
beginning of March 2014. We then calculated the intraday return
(equally weighted basis) and intraday turnover ratio on 25 March on
a strategy comprising long positions in the stocks in the quintile
with the highest factor values and short positions in the stocks in
the quintile with the lowest factor values. We did not take trading
costs into account. Analysis is based on historical data and does
not guarantee future performance.
Source: Nomura
09:00 09:30 10:00 10:30 11:00 11:30 12:00 12:30 13:00 13:30 14:00
14:30
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
(%) Intraday return
09:00 09:30 10:00 10:30 11:00 11:30 12:00 12:30 13:00 13:30 14:00
14:30 0
1
2
3
4
5
6
7
(%) Intraday factor return (lhs) and int raday turnover ratio
(rhs)
09:00 09:30 10:00 10:30 11:00 11:30 12:00 12:30 13:00 13:30 14:00
14:30 0
1
2
3
4
5
6
7
8
-1.4
-1.2
-1
-0.8
-0.6
-0.4
-0.2
0
The selling pressure concentrated on E/P-based factors and other
specific investment
styles that we saw in the Japanese equity market on 24 March was
reminiscent of the
"quant headwind" in August 2007 4 , when the performance of similar
factor strategies
plummeted simultaneously for a short period of time in a number of
markets in the US,
Europe, Japan, and Asia. Below, we set out our analysis of whether
what happened on
24–25 March was due to the unwinding of positions by global funds
or simply due to
domestic factors (such as domestic fund cancellations). We conclude
that what
happened on 24–25 March was linked to the unwinding of positions by
domestic pension
funds. This is because we were unable to confirm a correlation in
global factor
performance during the period around 24 March, unlike for August
2007, and there was
also a marked deterioration in the performance of domestic pension
portfolios in March,
as performance data for Japanese equity investments by investor
category show.
Even if the unwinding of positions by Japanese pension funds was
the direct cause of
what happened on 24–25 March, as we suspect, it is difficult to
pinpoint the exact
reasons for this. However, we need to bear in mind that the
dissolution of EPFs is about
to get fully under way. We cannot rule out the possibility that
what happened on 24–25
March was related to EPF dissolution, but whether or not this was
the case, a risk clearly
remains that a similar thing will happen again. This is because EPF
dissolution is
scheduled to take place over the next five years, and is likely to
have multiple peaks.
Taking a positive view, we could see the recent spate of position
unwinding as "disaster
training" in preparation for the risk of further unwinding of
pension fund positions in the
future. We would therefore like to stress the importance of
avoiding stocks subject to
herding by active funds. For example, assuming there is a high risk
of unwinding for
stocks in which only domestic pension funds are substantially
overweight, we think
avoiding these stocks is likely to be an effective strategy. This
kind of strategy would
have performed well from mid-January 2014 onward, and would have
generated positive
returns, without suffering any negative impact from the spate of
unwinding on 24 March
2014.
2.1 No sign of a connection with unw inding by global funds
We now compare the recent selling pressure thought to be due to the
unwinding of
positions by active funds with what happened in August 2007 in
order to investigate
whether th