hsbc risk 282506
TRANSCRIPT
-
8/3/2019 hsbc risk 282506
1/38
FX Quantitative Strategy
November 2010
Risk on risk off: the full story
Currency
Quant Special
Disclosures and Disclaimer This report must be read with the disclosures and analyst
certifications in the Disclosure appendix, and with the Disclaimer, which forms part of it
-
8/3/2019 hsbc risk 282506
2/38
1
Currency
FX Quantitative Strategy
10 November 2010
abc
Overview
In recent months, we have been analysing the correlations of returns across financial assets to understand
how the currently dominant market paradigm of risk on risk off developed. We have shown how
correlations between financial asset returns have intensified since the onset of the credit crisis and how
nearly all assets are now driven by a single, binary recovery factor. The market either believes that we are
on the road to recovery risk on; or that we are not risk off. We also argued that risk on risk off could
be the dominant market theme for some time as correlations are unlikely to fall back until the globaleconomic recovery is much better established.
Understanding the risk on risk off paradigm is important to all financial market participants because its
dominance means that portfolios may be less diversified than imagined and risks may therefore be higher
than desired. It also means that the search for relative value in markets is much harder as returns are
dominated by a single factor rather than by the nuances of individual market dynamics.
To provide the full story of the birth and growth of the risk on risk off paradigm, in this Quant Special
we bring together all of our prior analysis and extend it to give a more complete picture of the risk on
risk off phenomenon. In particular:
1 We introduce the new HSBC Risk On Risk Off (RORO) index to measure the extent to which the
risk on risk off phenomenon is driving markets. The index is indicative of the strength of market-
wide correlations and is currently at extremely high levels.
We will continue to track this index over the coming weeks in our recently revamped publication
HSBC Risk Indices. The index will enable us to identify when correlations decrease and markets
return to normal conditions.
2 We indicate the assets that are most strongly driven by the risk on risk off phenomenon and we
identify baskets of assets that provide diversified exposure to this factor.
3 We extend the history of the correlation heat maps back to 1990 to demonstrate the unprecedented
nature of the current situation.
4 We construct a risk on versus risk off index to identify whether the market is in a risk on or a risk
off state on a particular day.
-
8/3/2019 hsbc risk 282506
3/38
2
Currency
FX Quantitative Strategy
10 November 2010
abc
Contents
This Quant Special updates and extends our three recent analyses:Risk on-risk off how a paradigm is
born, Currency Weekly, 2 August 2010,What will end the risk on-risk off paradigm?, Currency Weekly,
16 August 2010 and Are we risk on or risk off today?, Currency Weekly, 13 September 2010. The
document is split into three parts:
1. Risk on risk off: the birth of a paradigm pg 3
We trace the birth of the risk on risk off paradigm by constructing correlation heat maps for 50 financial
assets since 2005. We show how the correlation between financial asset returns has intensified since the onset
of the credit crisis and how the risk on risk off paradigm remains the dominant market theme today.
2. What will end the risk on risk off paradigm? pg 13
The wide range of assets that we include in the first part limits the time period that we can consider. In the
second part, we therefore analyse a longer period of history by focusing on a smaller subset of assets. In
this section, we introduce the HSBC Risk On Risk Off (RORO) index and track its evolutions from
1990 up until today. We also try to identify the circumstances under which the current dominance of the
risk on risk off paradigm may start to fade. We conclude that it may dominate at least until the global
recovery is much more secure, which means it may be a crucial market feature for many months to come.
3. Are we risk on or risk off today? pg 21
We construct a risk on versus risk off index which tries to identify whether markets are currently in a
risk on or a risk off state.
The risk on risk off movie trilogy
As an aid to understanding the risk on risk off paradigm we attach links to three video clips. Clip 1
shows the evolution of correlation heat maps for 50 assets from 2005. Clip 2shows heat maps for 34
assets since 1990 and the risk on risk off index over that period.Clip 3is an interview with Stacy
Williams, Head of FX Quantitative Strategy at HSBC carried out by Risk magazine on the implications
of the risk on risk off phenomenon.
Video Clip 1: Click here to view video
Video Clip 2: Click here to view video
Video Clip 3: Click here to view video
https://www.research.hsbc.com/midas/Res/RDV?p=pdf&key=CjLQyHQJjB&n=274256.PDFhttps://www.research.hsbc.com/midas/Res/RDV?p=pdf&key=CjLQyHQJjB&n=274256.PDFhttps://www.research.hsbc.com/midas/Res/RDV?p=pdf&key=CjLQyHQJjB&n=274256.PDFhttps://www.research.hsbc.com/midas/Res/RDV?p=pdf&key=CjLQyHQJjB&n=274256.PDFhttps://www.research.hsbc.com/midas/Res/RDV?p=pdf&key=CjLQyHQJjB&n=274256.PDFhttps://www.research.hsbc.com/midas/Res/RDV?p=pdf&key=x7H6JVpPCP&n=275678.PDFhttps://www.research.hsbc.com/midas/Res/RDV?p=pdf&key=x7H6JVpPCP&n=275678.PDFhttps://www.research.hsbc.com/midas/Res/RDV?p=pdf&key=x7H6JVpPCP&n=275678.PDFhttps://www.research.hsbc.com/midas/Res/RDV?p=pdf&key=x7H6JVpPCP&n=275678.PDFhttp://research.uk.hibm.hsbc/midas/Res/RDV?p=pdf&$sessionid$=IPOt2HQvl15GiI4mGrAMZm5&key=ND17aVAYbW&n=277989.PDFhttp://research.uk.hibm.hsbc/midas/Res/RDV?p=pdf&$sessionid$=IPOt2HQvl15GiI4mGrAMZm5&key=ND17aVAYbW&n=277989.PDFhttp://research.uk.hibm.hsbc/midas/Res/RDV?p=pdf&$sessionid$=IPOt2HQvl15GiI4mGrAMZm5&key=ND17aVAYbW&n=277989.PDFhttp://cache.cantos.com/flash/hsba-r001/hsba-r001.avihttp://cache.cantos.com/flash/hsba-r001/hsba-r001.avihttp://cache.cantos.com/flash/hsba-r010/HSBC_heatmaps_from1990v2.avihttp://cache.cantos.com/flash/hsba-r010/HSBC_heatmaps_from1990v2.avihttp://www.hedgefundsreview.com/hedge-funds-review/news/1735715/video-senior-quant-stacy-williams-hsbc-discusses-cross-asset-correlationhttp://www.hedgefundsreview.com/hedge-funds-review/news/1735715/video-senior-quant-stacy-williams-hsbc-discusses-cross-asset-correlationhttp://www.hedgefundsreview.com/hedge-funds-review/news/1735715/video-senior-quant-stacy-williams-hsbc-discusses-cross-asset-correlationhttp://cache.cantos.com/flash/hsba-r001/hsba-r001.avihttp://cache.cantos.com/flash/hsba-r010/HSBC_heatmaps_from1990v2.avihttp://cache.cantos.com/flash/hsba-r010/HSBC_heatmaps_from1990v2.avihttp://www.hedgefundsreview.com/hedge-funds-review/news/1735715/video-senior-quant-stacy-williams-hsbc-discusses-cross-asset-correlationhttp://www.hedgefundsreview.com/hedge-funds-review/news/1735715/video-senior-quant-stacy-williams-hsbc-discusses-cross-asset-correlationhttps://www.research.hsbc.com/midas/Res/RDV?p=pdf&key=x7H6JVpPCP&n=275678.PDFhttps://www.research.hsbc.com/midas/Res/RDV?p=pdf&key=CjLQyHQJjB&n=274256.PDFhttps://www.research.hsbc.com/midas/Res/RDV?p=pdf&key=CjLQyHQJjB&n=274256.PDFhttp://www.hedgefundsreview.com/hedge-funds-review/news/1735715/video-senior-quant-stacy-williams-hsbc-discusses-cross-asset-correlationhttp://www.hedgefundsreview.com/hedge-funds-review/news/1735715/video-senior-quant-stacy-williams-hsbc-discusses-cross-asset-correlationhttp://cache.cantos.com/flash/hsba-r010/HSBC_heatmaps_from1990v2.avihttp://cache.cantos.com/flash/hsba-r001/hsba-r001.avihttp://cache.cantos.com/flash/hsba-r010/HSBC_heatmaps_from1990v2.avihttp://cache.cantos.com/flash/hsba-r001/hsba-r001.avihttp://research.uk.hibm.hsbc/midas/Res/RDV?p=pdf&$sessionid$=IPOt2HQvl15GiI4mGrAMZm5&key=ND17aVAYbW&n=277989.PDF -
8/3/2019 hsbc risk 282506
4/38
3
Currency
FX Quantitative Strategy
10 November 2010
abc
1. Risk on risk off: the birth of a paradigm
Recent high correlations between asset classes have led the market to become obsessed with the idea of
risk on risk off. The concept of risk on risk off is based on the markets view of the future state of the
world: the market either believes that future prospects are good, in which case risk is on; or the market
believes that future prospects are bad, in which case risk is off. In this section, we show how we have
moved from sophisticated and diverse markets to the simple binary risk on risk off mantra that
dominates today. This polarisation implies a high degree ofsynchronization between the movements of
different assets and consequently a high degree of correlation. Within this risk on risk off framework,
the nuances between different assets have disappeared, which makes diversification extremely difficult.
Hot heat maps
Although there is widespread acceptance of the idea of risk on risk off, the extent to which this
paradigm currently dominates markets is perhaps underestimated. The strength of correlations between a
wide variety of assets from different markets is best illustrated through correlation heat maps a pictorial
representation that draws out the main correlation structures.
A changing world...
In recent years, we have seen some of the most dramatic changes in financial markets that have ever been
witnessed. In Chart 1 we demonstrate the magnitude of these changes since 2005 using the VIX volatility
index. In this section, we just focus on some specific time periods to give a flavour of how the
correlations have changed and how we have arrived at todays risk on risk off way of trading. However,
in video clip 1we show the full evolution of market correlations over this period.
...driven by events
We selected a range of crucial events since 2005 that highlight different correlation regimes for this
publication. In video clip 1, we roll this window continuously from 2005 to today. The heat maps demonstrate
that correlations and the way that risk should be handled have changed dramatically since 2005. This means
that even if a pair of assets is highly correlated over one time period, they are not necessarily also correlated
over a later period. It also shows how we have moved from sophisticated markets, where asset allocation and
relative value were important, to todays simplified world of risk on risk off.
The periods indicated by horizontal bars in Chart 1 correspond to the following market events and conditions:
I. Normal (2005 to mid 2006)
II. Normal but awareness of the potential sub-prime crisis (2006 to mid 2007)
III. Crisis warnings and early crisis events (Northern Rock) an increase in correlation
IV. Attempt to normalize following early crisis pre Lehman
V. Crisis and correlations intensify collapse of Lehman Brothers
VI. Crisis high point strengthening correlationsVII. Risk on risk off: high correlations between all markets
VIII. Risk on risk off persists
http://cache.cantos.com/flash/hsba-r001/hsba-r001.avihttp://cache.cantos.com/flash/hsba-r001/hsba-r001.avihttp://cache.cantos.com/flash/hsba-r001/hsba-r001.avihttp://cache.cantos.com/flash/hsba-r001/hsba-r001.avihttp://cache.cantos.com/flash/hsba-r001/hsba-r001.avihttp://cache.cantos.com/flash/hsba-r001/hsba-r001.avi -
8/3/2019 hsbc risk 282506
5/38
4
Currency
FX Quantitative Strategy
10 November 2010
abc
The correlation heat map
We are interested in analyzing the correlations between all asset classes, so we consider a broad range of
assets covering the full spectrum of markets see Appendix A for details of the 50 different assets. These
include developed and emerging market equities, government and corporate bonds, commodities, interest
rates, credit, and currencies.
Our objective is to analyze the correlations between every pair of assets. However, for large numbers of
assets the number of correlations quickly becomes very large so it can be difficult to discern patterns by
just looking at a block of indigestible numbers. To get around this problem, in our HSBC heat maps (see
Charts 29) we represent the matrix of correlations between pairs of assets as an image in which different
colours correspond to different correlation strengths.
Heat map explained
In a heat map each row and each column corresponds to an asset, and the elements of the map are
coloured according to the correlation between asset pairs.
Dark Blue strongnegative correlation
Green weaknegative correlation or uncorrelated
Yellow weakpositive correlation or uncorrelated
Dark Red strong positive correlation.
For example, the diagonal of each heat map shows the correlations between each asset and itself; since
each asset is necessarily perfectly correlated with itself, the diagonal is always dark red.
In Charts 2-9, we show correlation heat maps for each of the time windows highlighted in Chart 1. We
will describe each heat map in turn and discuss what the map implies about market correlations. We
would advise reading this and then running video clip 1which moves smoothly through time.
1. The VIX Index since September 2005
0
10
20
30
40
50
60
70
80
90
Sep-05 Mar-06 Sep-06 Mar-07 Sep-07 Mar-08 Sep-08 Mar-09 Sep-09 Mar-10
0
10
20
30
40
50
60
70
80
90
VIXBernanke sub-prime warning
Run on Northern Rock
Lehman files for bankruptcy
Fed cuts rate to historical low
Greek fiscal crisis intensifies
I
II
III
IV
V
VI
VII
VIII
Source: HSBC, Bloomberg
http://cache.cantos.com/flash/hsba-r001/hsba-r001.avihttp://cache.cantos.com/flash/hsba-r001/hsba-r001.avihttp://cache.cantos.com/flash/hsba-r001/hsba-r001.avi -
8/3/2019 hsbc risk 282506
6/38
5
Currency
FX Quantitative Strategy
10 November 2010
abc
I.Normal (2005 to mid 2006)
In Chart 2, we show correlations over our normal period covering part of 2005 and 2006. We see lots of
green and yellow, indicating that most assets are uncorrelated. Here relative value and diverse asset
allocation strategies work well.
Some parts of the chart show high positive correlations (deep red), which we highlight with circles. These
are markets that are always highly correlated; for example, the assets that we highlight with the larger
circle are all bond yields, which tend to be highly correlated with each other. The lower circle highlights a
group of equities and, of course, it is unsurprising that different equity markets are also reasonably highly
correlated. However, despite these high correlations, most of the heat map is green and yellow, which
implies that most assets either have very weak correlations or are uncorrelated. For example, in this chart
bond yields and equity markets are not highly correlated with each other.
The range of greens and yellows implies there are many separate forces in markets that are driving
different assets in a non-trivial way. This range of forces leads to many different behaviours and
consequently to a large number of uncorrelated assets a very different world, as we shall show, to the
one that we find ourselves in today.
2. I. Normal (2005 to mid 2006)
Source: HSBC, Bloomberg
-
8/3/2019 hsbc risk 282506
7/38
6
Currency
FX Quantitative Strategy
10 November 2010
abc
II.
Normal but awareness of potential sub-prime crisis (2006 to mid 2007)
The heat map in Chart 3 shows the correlations over a period immediately preceding Fed Chairman
Bernankes warning of the extent of the potential losses from sub-prime lending. At this point the crisis
has not hit, but there are some minor differences between Charts 2 and 3. For example, the correlations
between bonds and equities have increased slightly, which is highlighted by a slight yellowing of the
circled region in Chart 3. However, overall the two heat maps are extremely similar. This demonstrates
the stability of correlations from the end of 2005 until the start of the crisis period in July 2007.
3. II: Normal but awareness of potential sub-prime crisis (2006 to mid 2007)
Source: HSBC, Bloomberg
-
8/3/2019 hsbc risk 282506
8/38
7
Currency
FX Quantitative Strategy
10 November 2010
abc
III.
Crisis warnings and early crisis events (N. Rock) an increase in correlation
Chart 4 follows the major early events in the sub-prime crisis, such as the Bernanke warning and the run
on Northern Rock in September 2007.
The circle towards the top of Chart 4 on the left hand side highlights that over this period the correlations
between bond yields and equities increased, which implies a stronger relationship between these asset
classes. The rightmost oval highlights that assets that were previously uncorrelated with bonds and
equities are now starting to become correlated with them. For example, correlations between equities and
commodities increase. This provides an early indication that the market might be heading towards a state
in which it is driven by a much smaller number of forces.
The most striking change in Chart 4 is the increase in the extent of the dark blue region at the bottom and
on the top right-hand side of the heat map which indicate strong negative correlations. Some of the assets
in this region include the VIX volatility index, credit, and the so called safe haven currencies CHF,
USD and JPY. These negative correlations are a manifestation of the same effect as that leading to high
positive correlations: the same force is driving some assets up and other assets down. The significance of
Chart 4 lies in the increased polarization in correlations. There are significantly fewer uncorrelated assets,
which implies that a single dominant force is driving markets.
4. III: Crisis warnings and early crisis events (Northern Rock) an increase in correlation
Source: HSBC, Bloomberg
-
8/3/2019 hsbc risk 282506
9/38
8
Currency
FX Quantitative Strategy
10 November 2010
abc
IV.
Attempt to normalize following early crisis pre Lehman
Greens and yellows fight back against reds and blues
The heat map in Chart 5 shows the time window from October 2007 to September 2008 covering the
period after the early crisis events, but before some of the events that would shake markets later in the
crisis. The correlations over this period suggest that markets are attempting to normalize after the early
shocks and return to their pre-crisis state. For example, there are decreases in bond-equity correlations,
and the strong negative correlations between the VIX and most other assets soften. Some of the subtle
differences between assets of the same type also seem to reappear; for example, some of the differences
between US and Asian equities re-emerged.
During this period, there appears to have been a shift back towards a market driven by multiple forces,
which results in more uncorrelated green and yellow areas. This return to some sort of normality,
however, proved to be short lived. Following the collapse of Lehman Brothers there was another major
shift in correlations.
5. IV: Attempt to normalize following early crisis pre Lehmans
Source: HSBC, Bloomberg
-
8/3/2019 hsbc risk 282506
10/38
9
Currency
FX Quantitative Strategy
10 November 2010
abc
V.
Crisis and correlations intensify collapse of Lehman
Chart 6 shows that following the collapse of Lehman there was a sharp increase in correlations between
many assets. This rise was particularly strong for equities (the circled group of assets). The deep red in
the heat map for correlations between equities implies that there is little relative value within equity
markets during this period; i.e., although it might previously have been possible to find returns in, say,
Asian equities that were not available in US equities, these differences have disappeared. Whilst there
might previously have been subtle differences between the same types of asset, during the crisis these
differences vanished. You were either in equities or out of them, but playing one equity market against
another would not be a source of increased returns.
6. V: Crisis and correlations intensify collapse of Lehman
Source: HSBC, Bloomberg
-
8/3/2019 hsbc risk 282506
11/38
10
Currency
FX Quantitative Strategy
10 November 2010
abc
VI.
Crisis high point strengthening correlations
As the crisis intensified, the correlations between equities increased (Chart 7); in addition, the correlations
spread to other markets. In Chart 7, we highlight the increase in correlations between commodities and
equities, but there were similar increases in correlations between equities and bonds and between
commodities and bonds. At the same time, the negative correlations between all of these assets and the
group including the VIX and USD, JPY and CHF increased.
The polarization of correlations has reached its highest levels so far during this period, with very few
uncorrelated assets. This implies that the market is being driven by one major force and this really marks
the birth of the risk on risk off paradigm that envelops markets today.
7. VI. Crisis high point strengthening correlations
Source: HSBC, Bloomberg
-
8/3/2019 hsbc risk 282506
12/38
11
Currency
FX Quantitative Strategy
10 November 2010
abc
VII.
Risk on risk off: high correlations between all markets
Intermediate shades have been replaced by the extreme colours
From 2009 to the start of 2010, the degree of correlation between markets progressively increased until at
the beginning of 2010 most markets were highly correlated. This is illustrated in Chart 8 by the extent of
the deep red region and deep blue regions, indicating strong positive and negative correlations,
respectively, and the reduction in the green and yellow regions. The heat map has lost some of its colour
shading as the intermediate shades have been replaced by the extreme colours, which is consistent with a
shift in markets to the risk on risk off paradigm and a binary world. Within this paradigm, one either
believes that risk is on or risk is off, and that this single factor drives all markets. Relative value isextremely difficult to identify and finding uncorrelated assets is extremely difficult.
8. VII: Risk onrisk off high correlations between all markets
Source: HSBC, Bloomberg
-
8/3/2019 hsbc risk 282506
13/38
12
Currency
FX Quantitative Strategy
10 November 2010
abc
VIII.
Risk on risk off persists
Chart 9 shows that, nine months on from Chart 8, there has been very little change, even while economic
recovery has emerged. In fact, correlations have strengthened even further. This picture is in stark
contrast to Chart 2 which shows correlations for 2005-2006. During this early period, correlations were
only strong between the same types of assets, and a large number of assets were uncorrelated, which
implies that there were many complex forces driving markets. Today, this structure no longer exists: most
assets represent essentially the same trade, and the idea of risk on risk off dominates the market.
However, the changes between Chart 2 and Chart 9 demonstrate that, far from being static, correlations
are constantly evolving, with major changes triggered by specific market events. Therefore, although risk
on risk off is the most appropriate paradigm for describing the market today, this picture is unlikely to
persist forever. We have seen that, following the early events in the credit crisis, there was an attempt by
markets to normalize, but that further crises prevented this from happening. At some point the market will
attempt to normalize again and this normalization should be visible in its correlation structure. In the next
section we analyse the evolution of correlations over a longer time period to try to identify the
circumstance under which risk on risk off may start to fade.
Click the following URL to view Video Clip 1: http://cache.cantos.com/flash/hsba-r001/hsba-r001.avi
9. VIII: Risk on risk off persists
Source: HSBC, Bloomberg
http://cache.cantos.com/flash/hsba-r001/hsba-r001.avihttp://cache.cantos.com/flash/hsba-r001/hsba-r001.avi -
8/3/2019 hsbc risk 282506
14/38
13
Currency
FX Quantitative Strategy
10 November 2010
abc
2. What will end the risk on risk off paradigm?
In this section we address the issue of the circumstances under which the risk on risk off paradigm may
start to fade. To do this, we introduce the HSBC Risk On Risk Off (RORO) index to quantify the extent
to which risk on risk off dominates markets. The index is indicative of the strength of market-wide
correlations.
We use the RORO index to analyse the conditions under which correlations are strong or weak. Our
conclusions are as follows:
1 Correlations between asset classes appear to be on a long-term upward trend, which may reflect the
growing internationalisation of financial markets and the improvements in information technology.
We should not, therefore, expect correlations to fall back to levels seen in the mid-2000s.
2 Correlations rise during most, but not all, crisis periods and fall back once the crisis has passed.
3 Correlations tend to rise during weak macro-economic conditions, and fall back when growth
is stronger.
4 High correlations tend to be associated with high levels of volatility, and vice versa. However,
correlations have stayed high in recent months despite declines in volatility. This suggests a structural
change could be taking place in markets.
The analysis suggests that a weakening of the risk on risk off paradigm is likely only once macro
conditions are improved in a sustainable way. This implies that the paradigm will continue to dominate
the market for some considerable time. Even when the paradigm fades, it would be wrong to expect
correlations to fall back to pre-crisis levels given their long-term upward trend.
A summary measure
The heat maps shown in the first section give a comprehensive view of the correlations between a wide
range of financial assets over the past five years. However, to investigate the circumstances under which
the correlations may start to decline again, a single measure over a much longer history is required. Given
the lack of data availability for some of the assets over a longer period, we use a reduced set of 34 assets
to construct the RORO index.
The RORO index is constructed using principal component analysis (PCA) and is based on the rolling
correlations between the daily returns of 34 assets since 1990. We provide technical details of the
methodology used to construct the index in Appendix B. In essence though, the RORO index measures
the extent to which the risk on risk off phenomenon is driving markets. An increase in the index
indicates that risk on risk off has become more dominant. The index is indicative of the strength of
market-wide correlations: an increase in the index implies that correlations have increased across many
different assets. We will be publishing the RORO index on a weekly basis in our HSBC Risk Indicespublication. The RORO index can take values between zero and one. A low level of the index suggests a
heat map similar to that shown in Chart 2; a high level suggests a similar heat map to Chart 9.
-
8/3/2019 hsbc risk 282506
15/38
14
Currency
FX Quantitative Strategy
10 November 2010
abc
Heat map movie: the directors cutIn video clip 2we extend video clip 1and show the evolution of the heat maps and the RORO index over
the full period from 1990 to 2010. In order to cover a longer period of time, it is necessary to reduce the
number of assets that we consider from 50 to 34; however, we still provide a comprehensive overview of
all major markets. In the video we highlight a number of events that have shaken markets during the past
20 years. The video illustrates the unprecedented levels that correlations have reached across a range of
markets since the credit crisis, and the unique nature of the current market environment. To view this
video, click the following link:
Click the following URL to view Video Clip 2: http://cache.cantos.com/flash/hsba-r010/HSBC_heatmaps_from1990v2.avi
10. RORO index measuring correlation has been rising over time
0
0.1
0.2
0.3
0.4
0.5
Feb-90 Feb-92 Feb-94 Feb-96 Feb-98 Feb-00 Feb-02 Feb-04 Feb-06 Feb-08 Feb-10
0
0.1
0.2
0.3
0.4
0.5
RORO IndexIndex Index
Source: HSBC, Bloomberg
Chart 10 shows the RORO index since 1990 along with a linear trend line. Two points are immediately
clear. First, the current position shows the strongest correlations seen at any time over the past 20 years. It
is therefore not surprising that risk on risk off dominates. Second, there is a clear upward trend in the
index over the period, with higher highs and higher lows over time. This suggests a secular upward move
in correlations. This may be associated with the growing internationalisation of financial markets and
products over the period, and also with the improvements in information technology that allow significant
news events to be rapidly disseminated around the world.
In a crisis, correlations significantly increase...
In addition to showing a secular uptrend, the correlation index displays wide variations over time. The
key question is: under what circumstances do correlations rise and fall? Given the recent experience, it
might be expected that financial crises always lead correlations to become stronger. To investigate this,
we annotate the correlation index chart with the main international financial crises over the period
(Chart 11). In video clip 2we illustrate how correlations changed following a wider range of major
international crises and market events.
http://cache.cantos.com/flash/hsba-r010/HSBC_heatmaps_from1990v2.avihttp://cache.cantos.com/flash/hsba-r010/HSBC_heatmaps_from1990v2.avihttp://cache.cantos.com/flash/hsba-r001/hsba-r001.avihttp://cache.cantos.com/flash/hsba-r001/hsba-r001.avihttp://cache.cantos.com/flash/hsba-r010/HSBC_heatmaps_from1990v2.avihttp://cache.cantos.com/flash/hsba-r010/HSBC_heatmaps_from1990v2.avihttp://cache.cantos.com/flash/hsba-r010/HSBC_heatmaps_from1990v2.avihttp://cache.cantos.com/flash/hsba-r010/HSBC_heatmaps_from1990v2.avihttp://cache.cantos.com/flash/hsba-r010/HSBC_heatmaps_from1990v2.avihttp://cache.cantos.com/flash/hsba-r010/HSBC_heatmaps_from1990v2.avihttp://cache.cantos.com/flash/hsba-r001/hsba-r001.avihttp://cache.cantos.com/flash/hsba-r010/HSBC_heatmaps_from1990v2.avi -
8/3/2019 hsbc risk 282506
16/38
15
Currency
FX Quantitative Strategy
10 November 2010
abc
11. Not all crises see correlations rise
0
0.1
0.2
0.3
0.4
0.5
Feb-90 Feb-92 Feb-94 Feb-96 Feb-98 Feb-00 Feb-02 Feb-04 Feb-06 Feb-08 Feb-10
0
0.1
0.2
0.3
0.4
0.5
RORO IndexIndex Index
ERM Asia NASDAQ Credit Greece
Source: HSBC, Bloomberg
However, as can be seen from the chart, not all financial crises lead to a rise in correlations across
financial assets, and correlations can rise without a specific financial market crisis. The rise in
correlations in 1990/91 was associated with the Iraq invasion of Kuwait and its impact on oil and equity
prices. The ERM crisis (September 1992-June 1993) had little impact on overall correlations. This may be
because it was an exclusively European crisis and was focused on the FX markets. Correlations rose
ahead of the Tequila crisis in December 1994 when Mexico devalued, but tended to fall back in its
aftermath. The Asian and Russian crisis (June 1997-November 1998) and the dot-com crash (March
2000-November 2001) did see correlations rise as the crises developed. Both of these crises had a wider
global dimension than either the ERM or tequila crises and had bigger impacts on asset markets. There
was a significant decrease in correlations following the dot-com bubble.
The rise in correlations associated with the credit crisis which started in 2007 is, of course, the most
dramatic, and has the longest duration in our sample. Even when it was widely perceived that the crisis
was over in the second half of 2009, correlations did not fall back to pre-crisis levels, and they are nowagain at their highs. However, the movements in the correlation appear to be more volatile than before.
Recession and correlation
There does appear to be an association between weak macro conditions and rising correlations. Chart 12
shows the correlation index against the NBER identified periods of recession in the US (July 1990-March
1991, March 2001-November 2001, and December 2007-June 2010). In each of these three periods,
correlations were either rising or very high, and correlations tended to fall back once the recession was
over, although, after the 2001 recession, correlations did not start to fall back until the beginning of 2003.
The mechanism involved here may be that weak macro conditions are associated with monetary easing,
which usually means strong bond market performance and, after a lag, a recovery in equity markets. For aperiod of time equities and bonds move together, pushing correlations higher. At the same time, low
yields in the major markets encourage the establishment of carry trades, which tends to mean that high-
-
8/3/2019 hsbc risk 282506
17/38
16
Currency
FX Quantitative Strategy
10 November 2010
abc
yielding currencies move in line with bonds and equities. These effects may be particularly extreme at the
moment given the close to zero money rates in the US, Japan and Europe.
Once economic growth recovers in a sustainable way, there is likely to be more diversity in money rates,
in the speed of monetary tightening and in asset market returns. This should see correlations fall back,
although there is no sign of this happening as yet.
12. Recessions seem to be associated with high correlations
0
0.1
0.2
0.3
0.4
0.5
Feb-90 Feb-92 Feb-94 Feb-96 Feb-98 Feb-00 Feb-02 Feb-04 Feb-06 Feb-08 Feb-10
0
0.1
0.2
0.3
0.4
0.5
RORO IndexIndex Index
Recessions
Source: HSBC, Bloomberg
Correlation and volatility
Chart 13 shows the relationship between the correlation index and a moving average of the VIX index of
implied equity market volatility. Higher levels of volatility appear to be associated with higher degrees of
correlation. This is clearly also related to periods of financial crisis, when there are typically very large
movements in financial asset prices, which will push volatility higher.
In previous periods of high volatility, once the peak has past there has been a decline in correlations.
However, in the current situation we have already seen a substantial decline in implied volatility from itspeak and yet correlations are still at their highs. To some extent, this may be because volatility has
become a much more widely traded instrument in recent years and therefore volatility itself may have
become more volatile. It may also indicate that the markets are a long way from being convinced that the
crisis is truly over given the overhang of bank balance sheet and fiscal problems in the major economies.
Correlations are currently at unprecedented levels
This inverse relationship between correlation and volatility is what really marks out the recent crisis from
all previous crises. Although volatility has decreased significantly since its peak at the height of the crisis,
there has not been a corresponding reduction in correlations. Correlations between many different assets
continue to persist at extremely high levels. The current extent of market-wide correlations is entirelywithout precedent and means that portfolios may be less diversified than imagined and risks may
therefore be higher than desired. The similarity in the behaviour of many different asset classes also
-
8/3/2019 hsbc risk 282506
18/38
17
Currency
FX Quantitative Strategy
10 November 2010
abc
means that the search for relative value is much harder and that many of the nuances between different
asset classes no longer exist.
13. High correlations tend to be associated with high volatility
0
0.1
0.2
0.3
0.4
0.5
Feb-90 Feb-92 Feb-94 Feb-96 Feb-98 Feb-00 Feb-02 Feb-04 Feb-06 Feb-08 Feb-10
0
10
20
30
40
50
60
ROR O Index ( LHS) VIX (80 day m a, RHS)
RORO Index vs VIXIndex Index
Source: HSBC, Bloomberg
The risk on risk off factorGiven the continued dominance of the risk on risk off paradigm, a question of widespread interest is:
which assets are most strongly affected by this phenomenon? In order to answer this question, we identify
a risk on risk off factor which represents returns that are attributable to risk on risk off. We provide
details of the methodology that we use to calculate this factor in Appendix B.
We measure the extent to which different assets are affected by risk on risk off by calculating the
correlation between the asset returns and the risk on risk off factor. High positive or negative
correlations indicate that an asset is strongly affected by risk on risk off.
In Chart 14, we show correlations between the risk on risk off factor and the 34 assets that we use to
construct the RORO index over the 80 days up to 4 November. During this time period the Russell 2000index was most strongly correlated with risk on and the VIX volatility index was most strongly correlated
with risk off. Perhaps the most striking aspect of Chart 14 is the number of assets that are currently being
driven by risk on risk off. This is illustrated by the fact that most of the assets have strong positive or
negative correlations with the risk on risk off factor.
It is also worth noting the assets that are least affected by risk on risk off. Over the 80 days up to 4
November 2010 this was GBP and gold. The weak correlations between these assets and the risk on risk
off factor suggests that these assets might represent opportunities for diversification.
-
8/3/2019 hsbc risk 282506
19/38
18
Currency
FX Quantitative Strategy
10 November 2010
abc
14. Asset correlations with the risk on risk off factor
Source: HSBC, Bloomberg
In Chart 15 we show the correlation heat map for the 80 days up to 4 November 2010. This brings theanalysis that we described inRisk on-risk off how a paradigm is born, Currency Weekly, 2 August
2010, up-to-date. The recent heat map is dominated by deep reds (indicating high positive correlations)
and deep blues (indicating high negative correlations), which indicates that a single factor is driving
markets. There are only small regions of green and yellow (indicating small correlations/uncorrelated
assets). Chart 15 therefore highlights the extent to which risk on risk off continues to dominate markets.
This single recovery factor continues to drive the behaviour of nearly all markets.
Which basket of assets is most correlated with risk on risk off?
The strengths of the correlations between individual assets and the risk on risk off factor identify the
optimal assets to trade if one has a view on whether the market is risk on or risk off. However, in orderto gain some diversification when trading risk on risk off, it is better to trade a basket of assets. This
reduces ones exposure to risks associated with particular assets. To identify the baskets that offer the best
exposure to the risk on risk off factor, we run optimizations to find the asset weightings that maximize
the correlation between a basket and the factor.
Strongly risk on Strongly risk offUncorrelated with risk on risk off
https://www.research.hsbc.com/midas/Res/RDV?p=pdf&key=CjLQyHQJjB&n=274256.PDFhttps://www.research.hsbc.com/midas/Res/RDV?p=pdf&key=CjLQyHQJjB&n=274256.PDFhttps://www.research.hsbc.com/midas/Res/RDV?p=pdf&key=CjLQyHQJjB&n=274256.PDFhttps://www.research.hsbc.com/midas/Res/RDV?p=pdf&key=CjLQyHQJjB&n=274256.PDFhttps://www.research.hsbc.com/midas/Res/RDV?p=pdf&key=CjLQyHQJjB&n=274256.PDF -
8/3/2019 hsbc risk 282506
20/38
19
Currency
FX Quantitative Strategy
10 November 2010
abc
In Table 16, we show the baskets of three instruments from particular asset classes that have the highest
correlations with the risk on. For example, an equity basket containing the S&P, the Russell 2000 and the
FTSE 100 indices with weights of 31%, 15%, and 54%, respectively, has the highest correlation of the
different equity baskets. Although this basket has the highest correlation with the risk on risk off factor,
however, a basket in which the S&P is replaced by the Dow Jones index has a similarly high correlation.
There are therefore several combinations of assets that produce almost equal exposure to risk on risk off.
Table 16 also shows that baskets in which the three assets are equally weighted have similar correlations
with the risk on risk off factors as baskets with weights that maximize the correlations. For example, in
the case of the basket containing S&P, Russell 200, and the FTSE 100, the optimally weighted basket has
a correlation of 0.97 with the risk on risk off factor, whereas the equally-weighted basket has a
correlation of 0.95. Given the similarity of the correlations for the two baskets, one can argue that it might
not be worth worrying about the optimal weightings since it is possible to gain good exposure to risk on
risk using an equally weighted basket. The advantage of using equal weights is that the basket is less
sensitive to any idiosyncrasies associated with particular assets.
15. Heat map showing correlationsover the 80 days up to 4 November 2010
Source: HSBC, Bloomberg
-
8/3/2019 hsbc risk 282506
21/38
20
Currency
FX Quantitative Strategy
10 November 2010
abc
16. Basket to gain exposure to risk on
Asset 1 Asset 2 Asset 3 Weight 1 Weight 2 Weight 3 Correlation Equal weightcorrelation
Equities
S&P Russell 2000 FTSE 100 0.31 0.15 0.54 0.970 0.954Dow Jones Russell 2000 FTSE 100 0.28 0.198 0.53 0.969 0.955
S&P FTSE 100 DAX 0.46 0.45 0.09 0.969 0.970
Bonds
US 10yr yield Germany 10yr yield Canada 10yr yield 0.06 0.34 0.601 0.860 0.840UK 10yr yield France 10yr yield Canada 10yr yield 0.13 0.19 0.686 0.839 0.792US 10yr yield France 10yr yield Canada 10yr yield 0.09 0.28 0.639 0.839 0.819
Commodities
Oil Copper Heating oil 0.29 0.28 0.43 0.8230 0.828
Soybean Copper Heating oil 0.17 0.27 0.55 0.829 0.809Wheat Copper Heating oil 0.07 0.30 0.63 0.828 0.704
Currencies
AUDJPY CADJPY NZDJPY 0.22 0.51 0.26 0.825 0.823
Source: HSBC, Bloomberg
Identifying the weightings of currency baskets for trading the risk on risk off is more complicated than
the other asset classes because one can be simultaneously long and short several different currencies
against each other. The correlations in Chart 14, however, indicate that one straightforward way to gain
exposure to risk on risk off is to go long some combination of currencies against the JPY, since the JPY
is by far the strongest risk off currency. For example, a basket that is long AUD, CAD, and NZD against
the JPY with weights of 22%, 51%, and 26%, respectively, has a correlation of 0.83 with the risk on risk off factor. An equally weighted basket has a very similar correlation.
When will risk on risk off end?
The analysis presented here suggests that correlations between returns in different financial assets tend to be
high in periods of financial crisis, in weak macroeconomic conditions, and when market volatility is high. Over
the past 20 years correlations have tended to fall back once the crisis is over, growth recovers, and volatility
falls back. However, there are clear signs of a rising trend in correlation, and it is striking that correlations are
currently still at their highs despite falls in levels of actual and implied market volatility. This may reflect
market concern that the crisis is not yet truly over and fears that weak macroeconomic conditions will remain
for a protracted period, or it may be that high correlations are now the new normal in financial markets In
either case, it would be wrong to expect correlations to fall for some time. Given this, we have also highlighted
the assets that can be used to gain exposure to the risk on risk off factor.
-
8/3/2019 hsbc risk 282506
22/38
21
Currency
FX Quantitative Strategy
10 November 2010
abc
3. Are we risk on or risk off today?
Identifying risk on risk off as the dominant paradigm is important, but it is equally important to try to
assess when the market switches from risk on to risk off and vice versa. As an aid to this, we construct
indices that measure the proportion of assets which are moving in a risk on or risk off fashion. We find
that while much of the year from March 2009 to March 2010 can be characterised as risk on, since April
we have largely been in a risk off environment with May and August being strongly risk off months.
Focussing on FX alone shows an even stronger contrast between the strongly risk on period up until March ofthis year and the risk off period since April. After a strongly risk off August, we seem to be tentatively back to
risk on since September, but we are still a long way from the risk on environment of 2009. Those expecting
risk on to continue should consider buying NZD, MXN and AUD against JPY, CHF and USD, looking to
reverse this should risk off again come to dominate the market.
Risk on versus risk off
The risk on versus risk off index measures the daily returns of 34 assets relative to risk on or risk off. For
example a positive equity market return is taken as risk on whereas a positive return for the yen is taken
as risk off. For each asset, risk on returns are assigned plus one, and risk off returns are assigned minus
one. The index is simply the average of these numbers across the assets. If all assets move in a risk ondirection, the index is +1; if all assets move in a risk off direction, the index is -1.
Chart 17 shows the index on a daily basis since January 1990. It is clear that there has been an increase in
the magnitude of the index over this period with more large positive and negative values today than in
earlier periods. The high proportion of times that the index reads +1 or -1 in recent years illustrates the
close correlation between assets today.
The other striking feature of the chart is the frequency with which the markets shift from full risk on to
full risk off. This implies a lack of conviction on long-term trends in the market, and also suggests that
volatility will remain relatively elevated.
-
8/3/2019 hsbc risk 282506
23/38
22
Currency
FX Quantitative Strategy
10 November 2010
abc
17. Markets move together an increasingly high proportion of the time
Daily Risk On vs Risk Off Indicator
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
Jan-90 Jan-92 Jan-94 Jan-96 Jan-98 Jan-00 Jan-02 Jan-04 Jan-06 Jan-08 Jan-10
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
Index Index
Source: HSBC, Bloomberg
Given the way the index oscillates between +1 and -1 it is difficult to interpret what it means for the state
of the risk on risk off paradigm. In order to make this clearer, Chart 18 shows the cumulative index over
the period. A rising index shows markets moving in a risk on way and a falling index shows markets
moving in a risk off way.
18. Markets have been in a mostly risk off mode since April 2010
Cumulative Risk On vs Risk Off Indicator
-20
-10
0
10
20
30
40
50
Jan-90 Jan-92 Jan-94 Jan-96 Jan-98 Jan-00 Jan-02 Jan-04 Jan-06 Jan-08 Jan-10
-20
-10
0
10
20
30
40
50
Cumulative Risk On vs Risk Off Indicator (LHS) 80 day ma (RHS)
Index Index
Risk on
Risk off
Source: HSBC, Bloomberg
As can be seen from the chart, most of the period from March 2009 to the beginning of April 2010 can be
described as risk on, although there were a number of significant reversals along the way. Since April, themarkets have been dominated by risk off.
-
8/3/2019 hsbc risk 282506
24/38
23
Currency
FX Quantitative Strategy
10 November 2010
abc
In Chart 18, we also show an 80 day moving average. If this is rising then the market can be described as
in risk on mode, and if it is falling then the market can be described as being in a risk off mode. Using this
criterion, the market has been in risk off mode since 19 August.
Another way of describing the market state is to look at the performance of the index by calendar month. On
this basis, the strongest risk on months were March and December 2009, and the strongest risk off months have
been May and August 2010 (Chart 19). In May, the dominant market driver was sovereign risk, with the Greek
crisis raising questions about government finances in several European countries. The factors driving the risk
off moves in August were discussed inThe holiday is over, Currency Weekly, 6 September 2010. Since the
beginning of September we have seen a return to risk on.
19. May and August 2010 were strongly risk off months
Monthly Risk On vs Risk Off Indicator
-5.0
-4.0
-3.0
-2.0
-1.0
0.0
1.0
2.0
3.0
4.0
5.0
Jan-08 Apr-08 Jul-08 Oct-08 Jan-09 Apr-09 Jul-09 Oct-09 Jan-10 Apr-10 Jul-10 Oct-10
-5.0
-4.0
-3.0
-2.0
-1.0
0.0
1.0
2.0
3.0
4.0
5.0
Index Index
Risk on
Risk off
Source: HSBC, Bloomberg
The movements in the risk on versus risk off index to some extent mirror the movements of equity
markets over the past two years. Chart 20 shows the index against the S&P500 since January 2008, and
for much of the period the movements are fairly similar. This is not too surprising as the S&P is still the
benchmark risk asset and the high levels of market correlation mean that a positive day for the S&P will
probably be associated with a positive day for most risk assets and a negative day for most safe haven
assets such as treasuries and the yen.
https://www.research.hsbc.com/midas/Res/RDV?p=pdf&key=W0V9MJj077&n=277485.PDFhttps://www.research.hsbc.com/midas/Res/RDV?p=pdf&key=W0V9MJj077&n=277485.PDFhttps://www.research.hsbc.com/midas/Res/RDV?p=pdf&key=W0V9MJj077&n=277485.PDFhttps://www.research.hsbc.com/midas/Res/RDV?p=pdf&key=W0V9MJj077&n=277485.PDFhttps://www.research.hsbc.com/midas/Res/RDV?p=pdf&key=W0V9MJj077&n=277485.PDF -
8/3/2019 hsbc risk 282506
25/38
24
Currency
FX Quantitative Strategy
10 November 2010
abc
20. Risk index and the S&P move together
Cumulativ e Risk On vs Risk Off Indicator v ersus S&P
28
30
32
34
36
38
40
Jan 08 Apr 08 Jul 08 Oct 08 Jan 09 Apr 09 Jul 09 Oct 09 Jan 10 Apr 10 Jul 10 Oct 10
Index
600
700
800
900
1000
1100
1200
1300
1400
1500Cumulativ e Risk On v s Risk Off Indicator (LHS) S&P 500 (RHS)
Index
Source: HSBC, Bloomberg
Since the recovery peak in the S&P in April, there looks from the chart as if there has been more divergence
between the equity performance and the risk on versus risk off index performance. However, this is somewhat
misleading (a result of the arbitrary scales on the chart) and in fact the correlation between daily moves in the
index and daily moves in the S&P has actually risen from 60% to 78% since the S&P peak. This may be
because the sovereign credit concerns that developed in May have further intensified the risk on risk off
paradigm such that risk assets performance is now even more closely aligned.
Risk on risk off in the FX market
Focussing in on the FX market, the same analysis can be carried out by looking at the returns of risk on
currencies (we have used AUD, NZD, MXN, ZAR, KRW, and INR) compared with the risk off currencies
(JPY and CHF). Chart 21 shows the result of calculating the same cumulative risk on versus risk off index for
these eight currencies since the beginning of 2009, along with a 20 day moving average.
-
8/3/2019 hsbc risk 282506
26/38
25
Currency
FX Quantitative Strategy
10 November 2010
abc
21. Risk on in the FX market turned to risk off in April
FX Risk On vs Risk Off indicator
-5
0
5
10
15
20
25
Dec 08 Mar 09 Jun 09 Sep 09 Dec 09 Mar 10 Jun 10 Sep 10
-5
0
5
10
15
20
25
Index Index
Risk on
Risk off
Source: HSBC, Bloomberg
The results for the FX market are somewhat similar to those obtained for the wider range of assets,
though the risk on period from March 2009 to April 2010 had fewer reversals. As with the wider asset
markets, risk off came to dominate from May 2010.
22. Strong risk on environment in 2009 followed by risk off since May 2010
FX Risk On v s Risk Off Index By Month
-5
0
5
10
15
20
Jan-09 Apr-09 Jul-09 Oct-09 Jan-10 Apr-10 Jul-10 Oct-10
-5
0
5
10
15
20Index Index
Source: HSBC, Bloomberg
In terms of monthly performance, risk on was in place for five consecutive months from March 2009, and
there was not a risk off month until January 2010 (Chart 22). May, June and August 2010 were all risk off
months and September saw a tentative return to risk on.
-
8/3/2019 hsbc risk 282506
27/38
26
Currency
FX Quantitative Strategy
10 November 2010
abc
The performance of the currency index is closely associated with the performance of a traditional FX
carry basket of high yielding currencies against low yielding currencies (Chart 23). This reflects the fact
that the main risk off currencies (JPY and CHF) have rates close to zero, whereas the main risk on
currencies (such as AUD and ZAR) have relatively high interest rates. A risk on period is associated with
strong carry performance and vice versa.
Carry on carry off
The carry trade is the closest parallel that the FX market has to equity beta. Historically, carry returns
have only been weakly correlated with equity returns and consequently provided a diversified source of
profits for investors. However, in a recent piece (Carry on carry off, Currency Weekly, 8 November
2010) we demonstrated that since the credit crisis this independence has broken down. Chart 23 shows
that carry is now driven by the risk on risk off factor and, as a result, carry returns are strongly
positively correlated with equity returns. In prevailing market conditions, speculative traders who engage
in the FX carry trade are therefore not exposed to the carry beta in the normal way. Instead, they are
simply exposed to the risk on risk off phenomenon.
23. Risk on versus risk off closely associated with carry returns
FX Risk On vs Risk Off Indicator and Carry Basket
-5
0
5
10
15
20
Dec 08 Mar 09 Jun 09 Sep 09 Dec 09 Mar 10 Jun 10 Sep 1090
95
100
105
110
115
120
125
130
135
Risk on vs risk off indicator (LHS)
Carry basket total return (RHS)
Index Index
Source: HSBC, Bloomberg
Which currencies do we buy?
If we can identify whether the market is in a risk on or a risk off mode, which currencies will perform
best and worst? In order to help answer this question we have looked at the best and worst performers
among the major currencies during the strongest risk on and risk off periods over the past two years. The
results are shown in Table 24.
http://www.research.hsbc.com/midas/Res/RDV?ao=20&key=ZZM6l0p07x&n=282876.PDFhttp://www.research.hsbc.com/midas/Res/RDV?ao=20&key=ZZM6l0p07x&n=282876.PDFhttp://www.research.hsbc.com/midas/Res/RDV?ao=20&key=ZZM6l0p07x&n=282876.PDFhttp://www.research.hsbc.com/midas/Res/RDV?ao=20&key=ZZM6l0p07x&n=282876.PDF -
8/3/2019 hsbc risk 282506
28/38
27
Currency
FX Quantitative Strategy
10 November 2010
abc
24. Currency performance in risk on and risk off periods
___________________Risk on ___________________ __________________Risk off___________________Dates 10 Mar 10 Apr 09 3 Sep 16 Oct 09 3 May 29 Jun 10 9 Aug 31 Aug 10
Best performers1 MXN NZD JPY CHF2 NZD AUD CHF JPY3 ZAR BRL USD USD
Worst performers1 JPY USD NOK NOK2 USD GBP AUD MXN3 CHF JPY KRW NZD
Source: HSBC, Bloomberg
Not surprisingly, AUD, NZD and MXN appear as both best performers in risk on periods and worstperformers in risk off periods. Equally, JPY and CHF are best performers in risk off periods and worst
performers in risk on periods. It should also be noted that the USD joins JPY and CHF in this.
The puzzling result is that the NOK was the worst performing currency in both the risk off periods earlier
this year. Given the very strong economic and financial fundamentals in Norway, we would have
expected it to perform much better in risk off periods, and would be reluctant to sell it should the market
move into risk off mode again.
The best currency selection for risk on would probably be long NZD, MXN and AUD against JPY, CHF
and USD with this being reversed during risk off periods.
Summary
The risk on risk off paradigm continues to dominate financial market performance with correlations
between asset returns at historically high levels. In the year to April 2010 signs of a global economic
recovery meant that risk on dominated market performance with equities, credit and high yielding
currencies performing well. Since April sovereign credit concerns and signs of weakening recovery have
made risk off the dominant force with high quality sovereign bonds and safe haven currencies performing
strongly. May and August 2010 were particularly strong risk off months. Since the beginning of
September some better economic data das seen risk on make a tentative return, but we are still a long way
from the risk on environment of 2009. Those expecting risk on to continue should consider buying NZD,
MXN and AUD against JPY, CHF and USD, looking to reverse this should risk off again come todominate the market.
-
8/3/2019 hsbc risk 282506
29/38
28
Currency
FX Quantitative Strategy
10 November 2010
abc
Conclusion
The risk on risk off paradigm has dominated financial markets in recent months. Our analysis shows
how the correlation between returns in different assets has tended to rise over a long period of time and
how it has reached unprecedented levels since the start of the financial crisis. Most strikingly, there is
little sign of these correlations falling back even though the immediate crisis has passed and despite
significant falls in implied volatility in markets.
Our analysis suggests that the risk on risk off paradigm will remain important to the markets at least
until the global economic recovery is much more secure. This means it could well dominate the markets
for many months to come.
HSBC Risk Indices
We will continue to track HSBC Risk On Risk Off (RORO) index over the coming weeks in our
recently revamped publication HSBC Risk Indices. The regular publication of the index will make it
possible to see any early signs of the paradigm fading. This will be crucial in determining the true level of
risk being run by a portfolio, and also the environment in which market relative value may again beimportant in portfolio construction.
The HSBC Risk Indices document also contains the following indices for measuring risk appetite:
OPRA: Position-based risk appetite index
The Open Positions Risk Appetite (OPRA) index measures risk appetite based on the futures positions
held by speculative traders in contracts with varying degrees of risk.
MRAI: Price-based risk appetite index
The Market Risk Appetite Index (MRAI) measures risk appetite based on changes in the price and
volatility of several assets that are known to be strongly affected by the markets appetite for risk.
-
8/3/2019 hsbc risk 282506
30/38
29
Currency
FX Quantitative Strategy
10 November 2010
abc
Appendix A: Heat map analysis
In our analysis for 2005-2010, we calculate correlations over a time window that includes 50 weekly returns for
each asset.
The key to how insightful the heat maps can be is the order of the assets in the rows and columns. The
obvious approach would be to order the assets based on their type; for example, to place all government
bonds next to each other, all equities next to each other, and so on. However, in some circumstances
particular assets can be more correlated with assets from different markets than they are with assets fromthe same market. With this in mind, we use a different approach to the ordering. We determine the order
using an optimization procedure that places correlated assets in adjacent rows (or columns). As we show,
this technique results in blocks of highly correlated assets in the correlation heat maps that do not
necessarily correspond to assets from the same class.
In the following table, we list all of the assets that we consider.
Details of the assets used in the heat maps
Asset Asset Asset
1 US 10yr bond yields 18 DAX 35 AAA corporate bond yields2 UK 10yr bond yields 19 Hang Seng 36 AA corporate bond yields3 Japan 10yr bond yields 20 Sao Paolo SX 37 USD*4 EU 10yr bond yields 21 Singapore SX 38 GBP5 Norway 10yr bond yields 22 Johannesburg SX 39 JPY6 Sweden 10 yr bond yields 23 EM Asia equities 40 EUR7 Australia 10yr bond yields 24 EM LatAm equities 41 NOK8 Canada 10yr bond yields 25 VIX 42 SEK9 Brazil 2yr bond yields 26 Corporate credit - main 43 AUD10 Singapore 10yr bond yields 27 Corporate credit - high vol. 44 CAD11 SA 10yr bond yields 28 Corporate credit - senior financials 45 BRL12 Hong Kong 10yr bond yields 29 Copper 46 ZAR13 S&P 30 Gold 47 CHF14 Russell 2000 31 Oil 48 3m eurodollar15 FTSE 100 32 Natural gas 49 3m euribor
16 Nikkei 33 Soybean 50 3m euroyen17 Eurostoxx 50 34 Wheat
Source: HSBC, Bloomberg
*All currencies are trade weighted indices.
-
8/3/2019 hsbc risk 282506
31/38
30
Currency
FX Quantitative Strategy
10 November 2010
abc
Appendix B: Risk on risk off index
In our analysis of the period 2005-2010, we used weekly data to minimize any effects due to the different
trading hours for markets in different regions. However, using weekly data meant that each time window
covered a full year of data, which can make it difficult to discern short-term changes in correlation. With this in
mind, we also analysed the period 1990-2010 using daily data. Because we use daily data, we focus on markets
that have a large overlap in trading hours (Europe and North America and Asian currency markets). This
enables us to track correlations using shorter time windows.
The RORO Index takes the rolling correlations between the daily returns of the 34 assets listed in the
table below and combines them into a single index. We construct the index by using principal component
analysis (PCA) to decompose the 34 asset return time series into 34 principal components (PCs), which
are mutually uncorrelated variables that explain the observed asset returns. The first PC represents the
most important factor driving financial markets during a particular time period. In current market
conditions, this factor can be considered to represent risk on risk off. The proportion of the variance
explained by the first PC then provides an indication of the strength with which this paradigm dominates
markets. If the first PC dominates markets and explains a large proportion of the variance, this implies
that market-wide correlations are strong, which is a key feature of the risk on risk off paradigm. In this
scenario, this single factor is driving synchronized changes amongst many different markets; hence
correlations are high.
We define the RORO Index as the variance in market returns explained by the first PC. An increase in the
RORO Index implies an increase in market correlations, whereas a decrease implies that market
correlations have decreased. In constructing the index we focus on markets that have a large overlap in
trading hours (Europe and North America and Asian currency markets). This enables us to track correlations on
a daily basis without having to worry about the non-synchronicity of return time series.
We also consider correlations between the different assets and the risk on risk off factor. These are
the correlations between the different return time series and the first PC, and can also be considered to
provide an indication of the extent to which risk on risk off is driving different assets.
Details of the assets used in the RORO index
Equities Government bonds(10 year yields)
Corporate bonds(yields)
Currencies tradeweights indices)
Metals Other
S&P US AAA USD Gold VIXDow Jones Canada BAA EUR Silver OilNASDAQ UK CHF Copper Natural Gas
Russell 2000 Germany GBP Heating OilFTSE 100 France JPY Wheat
Euro Stoxx 50 AUD SoybeanDAX CAD Cotton
CAC 40 NZD
Source: HSBC, Bloomberg
-
8/3/2019 hsbc risk 282506
32/38
31
Currency
FX Quantitative Strategy
10 November 2010
abc
Notes
-
8/3/2019 hsbc risk 282506
33/38
32
Currency
FX Quantitative Strategy
10 November 2010
abc
Notes
-
8/3/2019 hsbc risk 282506
34/38
33
Currency
FX Quantitative Strategy
10 November 2010
abc
Notes
-
8/3/2019 hsbc risk 282506
35/38
34
Currency
FX Quantitative Strategy
10 November 2010
abc
Disclosure appendix
Analyst Certification
The following analyst(s), economist(s), and/or strategist(s) who is(are) primarily responsible for this report, certifies(y) that the
opinion(s) on the subject security(ies) or issuer(s) and/or any other views or forecasts expressed herein accurately reflect their
personal view(s) and that no part of their compensation was, is or will be directly or indirectly related to the specific
recommendation(s) or views contained in this research report: Stacy Williams, Daniel Fenn, Mark McDonald, Paul Mackel and
David Bloom
Important Disclosures
This document has been prepared and is being distributed by the Research Department of HSBC and is intended solely for the
clients of HSBC and is not for publication to other persons, whether through the press or by other means.
This document is for information purposes only and it should not be regarded as an offer to sell or as a solicitation of an offer
to buy the securities or other investment products mentioned in it and/or to participate in any trading strategy. Advice in this
document is general and should not be construed as personal advice, given it has been prepared without taking account of the
objectives, financial situation or needs of any particular investor. Accordingly, investors should, before acting on the advice,
consider the appropriateness of the advice, having regard to their objectives, financial situation and needs. If necessary, seek
professional investment and tax advice.
Certain investment products mentioned in this document may not be eligible for sale in some states or countries, and they may
not be suitable for all types of investors. Investors should consult with their HSBC representative regarding the suitability ofthe investment products mentioned in this document and take into account their specific investment objectives, financial
situation or particular needs before making a commitment to purchase investment products.
The value of and the income produced by the investment products mentioned in this document may fluctuate, so that an
investor may get back less than originally invested. Certain high-volatility investments can be subject to sudden and large falls
in value that could equal or exceed the amount invested. Value and income from investment products may be adversely
affected by exchange rates, interest rates, or other factors. Past performance of a particular investment product is not indicative
of future results.
Analysts, economists, and strategists are paid in part by reference to the profitability of HSBC which includes investment
banking revenues.
For disclosures in respect of any company mentioned in this report, please see the most recently published report on thatcompany available at www.hsbcnet.com/research.
* HSBC Legal Entities are listed in the Disclaimer below.
Additional disclosures
1 This report is dated as at 10 November 2010.2 All market data included in this report are dated as at close 09 November 2010, unless otherwise indicated in the report.3 HSBC has procedures in place to identify and manage any potential conflicts of interest that arise in connection with its
Research business. HSBC's analysts and its other staff who are involved in the preparation and dissemination of Researchoperate and have a management reporting line independent of HSBC's Investment Banking business. Information Barrierprocedures are in place between the Investment Banking and Research businesses to ensure that any confidential and/or
price sensitive information is handled in an appropriate manner.
-
8/3/2019 hsbc risk 282506
36/38
35
Currency
FX Quantitative Strategy
10 November 2010
abc
Disclaimer
* Legal entities as at 31 January 2010
'UAE' HSBC Bank Middle East Limited, Dubai; 'HK' The Hongkong and Shanghai Banking Corporation
Limited, Hong Kong; 'TW' HSBC Securities (Taiwan) Corporation Limited; 'CA' HSBC Securities (Canada)
Inc, Toronto; HSBC Bank, Paris branch; HSBC France; 'DE' HSBC Trinkaus & Burkhardt AG, Dusseldorf;
000 HSBC Bank (RR), Moscow; 'IN' HSBC Securities and Capital Markets (India) Private Limited, Mumbai;
'JP' HSBC Securities (Japan) Limited, Tokyo; 'EG' HSBC Securities Egypt S.A.E., Cairo; 'CN' HSBC
Investment Bank Asia Limited, Beijing Representative Office; The Hongkong and Shanghai Banking
Corporation Limited, Singapore branch; The Hongkong and Shanghai Banking Corporation Limited, Seoul
Securities Branch; The Hongkong and Shanghai Banking Corporation Limited, Seoul Branch; HSBC
Securities (South Africa) (Pty) Ltd, Johannesburg; 'GR' HSBC Pantelakis Securities S.A., Athens; HSBC
Bank plc, London, Madrid, Milan, Stockholm, Tel Aviv, 'US' HSBC Securities (USA) Inc, New York; HSBC
Yatirim Menkul Degerler A.S., Istanbul; HSBC Mxico, S.A., Institucin de Banca Mltiple, Grupo
Financiero HSBC, HSBC Bank Brasil S.A. - Banco Mltiplo, HSBC Bank Australia Limited, HSBC Bank
Argentina S.A., HSBC Saudi Arabia Limited.
Issuer of report
HSBC Bank plc
8 Canada Square, London
E14 5HQ, United Kingdom
Telephone: +44 20 7991 8888
Telex: 888866
Fax: +44 20 7992 4880
Website: www.research.hsbc.com
This document is issued and approved in the United Kingdom by HSBC Bank plc for the information of its Clients (as defined in the Rules of FSA) and those
of its affiliates only. If this research is received by a customer of an affiliate of HSBC, its provision to the recipient is subject to the terms of business in place
between the recipient and such affiliate. In Australia, this publication has been distributed by The Hongkong and Shanghai Banking Corporation Limited
(ABN 65 117 925 970, AFSL 301737) for the general information of its wholesale customers (as defined in the Corporations Act 2001). Where distributed
to retail customers, this research is distributed by HSBC Bank Australia Limited (AFSL No. 232595). These respective entities make no representations that
the products or services mentioned in this document are available to persons in Australia or are necessarily suitable for any particular person or appropriate in
accordance with local law. No consideration has been given to the particular investment objectives, financial situation or particular needs of any recipient.
The document is distributed in Hong Kong and Japan by The Hongkong and Shanghai Banking Corporation Limited and has been prepared for the New York
office of HSBC Bank USA, National Association. In Korea, this publication is distributed by either The Hongkong and Shanghai Banking Corporation
Limited, Seoul Securities Branch ("HBAP SLS") or The Hongkong and Shanghai Banking Corporation Limited, Seoul Branch ("HBAP SEL") for the general
information of professional investors specified in Article 9 of the Financial Investment Services and Capital Markets Act (FSCMA). This publication is not
a prospectus as defined in the FSCMA. It may not be further distributed in whole or in part for any purpose. Both HBAP SLS and HBAP SEL are regulated
by the Financial Services Commission and the Financial Supervisory Service of Korea.
Each of the companies listed above (the Participating Companies) is a member of the HSBC Group of Companies, any member of which may trade for its own
account as Principal, may have underwritten an issue within the last 36 months or, together with its Directors, officers and employees, may have a long or short
position in securities or instruments or in any related instrument mentioned in the document. Brokerage or fees may be earned by the Participating Companies or
persons associated with them in respect of any business transacted by them in all or any of the securities or instruments referred to in this document.
The information in this document is derived from sources the Participating Companies believe to be reliable but which have not been independently verified.
The Participating Companies make no guarantee of its accuracy and completeness and are not responsible for errors of transmission of factual or analytical
data, nor shall the Participating Companies be liable for damages arising out of any persons reliance upon this information. All charts and graphs are from
publicly available sources or proprietary data. The opinions in this document constitute the present judgement of the Participating Companies, which is
subject to change without notice.
This document is neither an offer to sell, purchase or subscribe for any investment nor a solicitation of such an offer. HSBC Securities (USA) Inc. accepts
responsibility for the content of this research report prepared by its non-US foreign affiliate. All US persons receiving and/or accessing this report and
intending to effect transactions in any security discussed herein should do so with HSBC Securities (USA) Inc. in the United States and not with its non-US
foreign affiliate, the issuer of this report. In Singapore, this publication is distributed by The Hongkong and Shanghai Banking Corporation Limited,
Singapore Branch for the general information of institutional investors or other persons specified in Sections 274 and 304 of the Securities and Futures Act
(Chapter 289) (SFA) and accredited investors and other persons in accordance with the conditions specified in Sections 275 and 305 of the SFA. This
publication is not a prospectus as defined in the SFA. It may not be further distributed in whole or in part for any purpose. The Hongkong and Shanghai
Banking Corporation Limited Singapore Branch is regulated by the Monetary Authority of Singapore. Recipients in Singapore should contact a "Hongkong
and Shanghai Banking Corporation Limited, Singapore Branch" representative in respect of any matters arising from, or in connection with this report. HSBC
Mxico, S.A., Institucin de Banca Mltiple, Grupo Financiero HSBC is authorized and regulated by Secretara de Hacienda y Crdito Pblico and Comisin
Nacional Bancaria y de Valores (CNBV). HSBC Bank (Panama) S.A. is regulated by Superintendencia de Bancos de Panama. Banco HSBC Honduras S.A. is
regulated by Comisin Nacional de Bancos y Seguros (CNBS). Banco HSBC Salvadoreo, S.A. is regulated by Superintendencia del Sistema Financiero
(SSF). HSBC Colombia S.A. is regulated by Superintendencia Financiera de Colombia. Banco HSBC Costa Rica S.A. is supervised by Superintendencia
General de Entidades Financieras (SUGEF). Banistmo Nicaragua, S.A. is authorized and regulated by Superintendencia de Bancos y de Otras Instituciones
Financieras (SIBOIF).
The document is intended to be distributed in its entirety. Unless governing law permits otherwise, you must contact a HSBC Group member in your home
jurisdiction if you wish to use HSBC Group services in effecting a transaction in any investment mentioned in this document. HSBC Bank plc is registered in
England No 14259, is authorised and regulated by the Financial Services Authority and is a member of the London Stock Exchange. (070905)
Copyright. HSBC Bank plc 2010, ALL RIGHTS RESERVED. No part of this publication may be reproduced, stored in a retrieval system, or transmitted,
on any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior written permission of HSBC Bank plc. MICA
(P) 142/06/2010 and MICA (P) 193/04/2010
[282506]
-
8/3/2019 hsbc risk 282506
37/38
abc
Global
David BloomGlobal Head of Currency+44 20 7991 5969 [email protected]
Asia
Richard Yetsenga+852 2996 6565 [email protected]
Perry Kojodjojo+852 2996 6568 [email protected]
Daniel Hui+852 2822 4340 [email protected]
United Kingdom
Paul Mackel
+44 20 7991 5968 [email protected] Williams
+44 20 7991 5967 [email protected]
Mark McDonald+44 20 7991 5966 [email protected]
Daniel Fenn+44 20 7991 5003 [email protected]
Mark AustinConsultant
United States
Robert Lynch
+1 212 525 3159 [email protected]
Clyde Wardle
+1 212 525 3345 [email protected]
Marjorie Hernandez+1 212 525 4109 [email protected]
Technical Analysis
Murray Gunn+44 20 7991 5384 murray,[email protected]
Precious Metals
James Steel
+1 212 525 6515 [email protected]
Global Currency Strategy Research Team
-
8/3/2019 hsbc risk 282506
38/38
Main Contributors
Mark McDonaldFX Quantitative StrategistHSBC Bank plc+44 20 7991 [email protected]
Mark is a quantitative FX strategist based in London. He joined HSBC in 2005. Before joining the company, he obtained a
DPhil from Oxford University, researching in collaboration with the HSBC FX Strategy team. Mark has an MPhys in Physics,
also from Oxford University.
Paul MackelDirector of Currency StrategyHSBC Bank plc+44 20 7991 [email protected]
Paul is senior currency strategist covering the G10 currency markets. He joined HSBC in June 2006 and is based in London.
Prior to joining the company, Paul worked in similar roles for other financial institutions. He is a regular contributor to the
FX strategy publications.
Stacy WilliamsHead of FX Quantitative StrategyHSBC Bank plc+44 20 7991 [email protected]
Stacy Williams is Head of FX Quantitative Strategy. His responsibilities include producing quantitative research, advising on
the development of currency overlay programs and the construction of bespoke hedging strategies. Stacy is also responsiblefor proprietary model trading, concentrating on models based on transactional flow information, high frequency price data
and economic activity data.
Daniel FennFX Quantitative StrategistHSBC Bank plc+44 20 7991 [email protected]
Dan is a quantitative FX strategist based in London. Before joining HSBC in 2009, Dan was studying for a PhD at the
University of Oxford, researching in collaboration with the HSBC FX Strategy team.
David BloomGlobal Head of FX ResearchHSBC Bank plc+44 20 7991 [email protected]
David is the Global Head of Foreign Exchange Strategy for HSBC. He has been with the Group since 1992. Before taking up
his current post, specialising in currencies and market strategies, David was the US economist for the Bank. He also has work
experience within equity markets and analysing the UK economy.