hsbc risk 282506

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


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


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


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


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


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


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)


8/3/2019 hsbc risk 282506

7/38

6

Currency


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)


8/3/2019 hsbc risk 282506

8/38

7

Currency


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


8/3/2019 hsbc risk 282506

9/38

8

Currency


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


8/3/2019 hsbc risk 282506

10/38

9

Currency


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


8/3/2019 hsbc risk 282506

11/38

10

Currency


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


8/3/2019 hsbc risk 282506

12/38

11

Currency


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


8/3/2019 hsbc risk 282506

13/38

12

Currency


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

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


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


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


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


10 November 2010

abc

11. Not all crises see correlations rise

0

0.1

0.2

0.3

0.4

0.5


0

0.1

0.2

0.3

0.4

0.5


ERM Asia NASDAQ Credit Greece


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


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


0

0.1

0.2

0.3

0.4

0.5


Recessions


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


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


0

10

20

30

40

50

60

ROR O Index ( LHS) VIX (80 day m a, RHS)

RORO Index vs VIXIndex Index


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


10 November 2010

abc

14. Asset correlations with the risk on risk off factor


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


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


8/3/2019 hsbc risk 282506

21/38

20

Currency


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


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


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


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


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


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


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


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


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


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


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


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


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


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


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


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


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


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


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


*All currencies are trade weighted indices.

8/3/2019 hsbc risk 282506

31/38

30

Currency


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


8/3/2019 hsbc risk 282506

32/38

31

Currency


10 November 2010

abc

Notes

8/3/2019 hsbc risk 282506

33/38

32

Currency


10 November 2010

abc

Notes

8/3/2019 hsbc risk 282506

34/38

33

Currency


10 November 2010

abc

Notes

8/3/2019 hsbc risk 282506

35/38

34

Currency


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


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


Clyde Wardle


Marjorie Hernandez+1 212 525 4109 [email protected]

Technical Analysis

Murray Gunn+44 20 7991 5384 murray,[email protected]

Precious Metals

James Steel


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.

hsbc risk 282506

Documents