HSBC Risk 282506

Quant Special
Currency
FX Quantitative Strategy November 2010
Risk on risk off: the full story
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
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 global economic 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 marketwide 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. 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.
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 2 shows heat maps for 34 assets since 1990 and the risk on risk off index over that period. Clip 3 is 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
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 of synchronization 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 1 we 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 correlations VII. Risk on risk off: high correlations between all markets VIII. Risk on risk off persists
abc
1. The VIX Index since September 2005

VIX VIII VII VI V IV III II I 90 80 70 60 50 40 30 20 10 0 Mar-06 Sep-06 Mar-07 Sep-07 Mar-08 Sep-08 Mar-09 Sep-09 Mar-10
Bernanke sub-prime warning

90 80 70 60 50 40 30 20 10 0 Sep-05
Run on Northern Rock Lehman files for bankruptcy Fed cuts rate to historical low Greek fiscal crisis intensifies
Source: HSBC, Bloomberg
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 strong negative correlation Green weak negative correlation or uncorrelated Yellow weak positive 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 1 which moves smoothly through time.
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)
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)
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
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
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
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
10
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 is extremely difficult to identify and finding uncorrelated assets is extremely difficult.
8. VII: Risk onrisk off high correlations between all markets
11
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
12
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. Correlations rise during most, but not all, crisis periods and fall back once the crisis has passed. Correlations tend to rise during weak macro-economic conditions, and fall back when growth is stronger. 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.
2 3
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 Indices publication. 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.
13
abc
Heat map movie: the directors cut

In video clip 2 we extend video clip 1 and 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
Index 0.5 RORO Index Index 0.5
0.4
0.4
0.3
0.3
0.2
0.2
0.1
0.1
0 Feb-90 Feb-92 Feb-94 Feb-96 Feb-98 Feb-00 Feb-02 Feb-04 Feb-06 Feb-08 Feb-10
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 2 we illustrate how correlations changed following a wider range of major international crises and market events.
14
abc
11. Not all crises see correlations rise

Index 0.5 ERM 0.4 Asia RORO Index NASDAQ Credit Greece 0.4 Index 0.5
0.3
0.3
0.2
0.2
0.1
0.1
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 now again 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 a period 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-
15
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
Index 0.5 Recessions 0.4 0.4 RORO Index Index 0.5
0.3
0.3
0.2
0.2
0.1
0.1
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 its peak 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 entirely without 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
16
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
Index 0.5 RORO Index v s VIX RORO Index (LHS) 0.4 VIX (80 day ma, RHS) 50 Index 60
40 0.3 30 0.2 20 0.1
10 0 Feb-92 Feb-94 Feb-96 Feb-98 Feb-00 Feb-02 Feb-04 Feb-06 Feb-08 Feb-10
0 Feb-90
The risk on risk off factor

Given 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 2000 index 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.
17
abc
14. Asset correlations with the risk on risk off factor
Strongly risk on
Uncorrelated with risk on risk off Strongly risk off
In Chart 15 we show the correlation heat map for the 80 days up to 4 November 2010. This brings the analysis that we described in Risk 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 order to 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.
18
abc
15. Heat map showing correlations over the 80 days up to 4 November 2010
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.
19
abc
16. Basket to gain exposure to risk on Asset 1 Equities S&P Dow Jones S&P Bonds US 10yr yield UK 10yr yield US 10yr yield Commodities Oil Soybean Wheat Currencies AUDJPY
Asset 2
Asset 3
Weight 1
Weight 2
Weight 3
Correlation
Equal weight correlation
Russell 2000 Russell 2000 FTSE 100 Germany 10yr yield France 10yr yield France 10yr yield Copper Copper Copper CADJPY
FTSE 100 FTSE 100 DAX Canada 10yr yield Canada 10yr yield Canada 10yr yield Heating oil Heating oil Heating oil NZDJPY
0.31 0.28 0.46 0.06 0.13 0.09 0.29 0.17 0.07 0.22
0.15 0.198 0.45 0.34 0.19 0.28 0.28 0.27 0.30 0.51
0.54 0.53 0.09 0.601 0.686 0.639 0.43 0.55 0.63 0.26
0.970 0.969 0.969 0.860 0.839 0.839 0.8230 0.829 0.828 0.825
0.954 0.955 0.970 0.840 0.792 0.819 0.828 0.809 0.704 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.
20
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 of this 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 on direction, 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.
21
abc
17. Markets move together an increasingly high proportion of the time

Index 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 Daily Risk On v s Risk Off Indicator Index 1.0 0.8 0.6 0.4 0.2 0.0 -0.2 -0.4 -0.6 -0.8 -1.0
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
Index 50 Cumulativ e Risk On v s Risk Off Indicator (LHS) 40 30 20 10 0 -10 -20 Jan-90 Jan-92 Jan-94 Jan-96 Jan-98 Jan-00 Jan-02 Jan-04 Jan-06 Jan-08 Jan-10 Risk on 80 day ma (RHS) Cumulativ e Risk On v s Risk Off Indicator Risk off Index 50 40 30 20 10 0 -10 -20
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, the markets have been dominated by risk off.
22
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 in The 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
Index 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 Risk off Risk on Monthly Risk On v s Risk Off Indicator Index 5.0 4.0 3.0 2.0 1.0 0.0 -1.0 -2.0 -3.0 -4.0 -5.0
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.
23
abc
20. Risk index and the S&P move together

Index 40 Cumulativ e Risk On v s Risk Off Indicator (LHS) 38 1300 36 1200 1100 34 1000 32 900 800 30 700 28 Jan 08 Apr 08 Jul 08 Oct 08 Jan 09 Apr 09 Jul 09 Oct 09 Jan 10 Apr 10 Jul 10 Oct 10 600 S&P 500 (RHS) 1400 Cumulativ e Risk On v s Risk Off Indicator v ersus S&P Index 1500
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.
24
abc
21. Risk on in the FX market turned to risk off in April
Index 25
FX Risk On v s Risk Off indicator
Index 25 Risk off
20 Risk on 15
20
15
10
10
-5 Dec 08
-5 Mar 09 Jun 09 Sep 09 Dec 09 Mar 10 Jun 10 Sep 10
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
Index 20 FX Risk On v s Risk Off Index By Month Index 20
15
15
10
10
-5 Jan-09 Apr-09 Jul-09 Oct-09 Jan-10 Apr-10 Jul-10 Oct-10
-5
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.
25
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
Index 20
FX Risk On v s Risk Off Indicator and Carry Basket Risk on v s risk off indicator (LHS)
Index 135 130 125 120
15
Carry basket total return (RHS)
10 115 110 5 105 0 100 95 -5 Dec 08

90 Mar 09 Jun 09 Sep 09 Dec 09 Mar 10 Jun 10 Sep 10
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.
26
abc
24. Currency performance in risk on and risk off periods Dates Best performers 1 2 3 Worst performers 1 2 3
___________________Risk on ___________________ 10 Mar 10 Apr 09 3 Sep 16 Oct 09 MXN NZD ZAR JPY USD CHF NZD AUD BRL USD GBP JPY
__________________ Risk off___________________ 3 May 29 Jun 10 9 Aug 31 Aug 10 JPY CHF USD NOK AUD KRW CHF JPY USD NOK MXN NZD
Not surprisingly, AUD, NZD and MXN appear as both best performers in risk on periods and worst performers 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 to dominate the market.
27
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 be important 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.
28
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 from the 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 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 US 10yr bond yields UK 10yr bond yields Japan 10yr bond yields EU 10yr bond yields Norway 10yr bond yields Sweden 10 yr bond yields Australia 10yr bond yields Canada 10yr bond yields Brazil 2yr bond yields Singapore 10yr bond yields SA 10yr bond yields Hong Kong 10yr bond yields S&P Russell 2000 FTSE 100 Nikkei Eurostoxx 50 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 Asset DAX Hang Seng Sao Paolo SX Singapore SX Johannesburg SX EM Asia equities EM LatAm equities VIX Corporate credit - main Corporate credit - high vol. Corporate credit - senior financials Copper Gold Oil Natural gas Soybean Wheat 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 Asset AAA corporate bond yields AA corporate bond yields USD* GBP JPY EUR NOK SEK AUD CAD BRL ZAR CHF 3m eurodollar 3m euribor 3m euroyen
All currencies are trade weighted indices.
29
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 S&P Dow Jones NASDAQ Russell 2000 FTSE 100 Euro Stoxx 50 DAX CAC 40
Government bonds (10 year yields) US Canada UK Germany France
Corporate bonds (yields) AAA BAA
Currencies trade weights indices) USD EUR CHF GBP JPY AUD CAD NZD
Metals Gold Silver Copper
Other VIX Oil Natural Gas Heating Oil Wheat Soybean Cotton
30
abc
Notes
31
abc
Notes
32
abc
Notes
33
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 of the 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 that company available at www.hsbcnet.com/research. * HSBC Legal Entities are listed in the Disclaimer below.
Additional disclosures
1 2 3 This report is dated as at 10 November 2010. All market data included in this report are dated as at close 09 November 2010, unless otherwise indicated in the report. 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 Research operate and have a management reporting line independent of HSBC's Investment Banking business. Information Barrier procedures 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.
34
abc
Disclaimer
* Legal entities as at 31 January 2010 Issuer of report 'UAE' HSBC Bank Middle East Limited, Dubai; 'HK' The Hongkong and Shanghai Banking Corporation HSBC Bank plc Limited, Hong Kong; 'TW' HSBC Securities (Taiwan) Corporation Limited; 'CA' HSBC Securities (Canada) 8 Canada Square, London Inc, Toronto; HSBC Bank, Paris branch; HSBC France; 'DE' HSBC Trinkaus & Burkhardt AG, Dusseldorf; E14 5HQ, United Kingdom 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 Telephone: +44 20 7991 8888 Investment Bank Asia Limited, Beijing Representative Office; The Hongkong and Shanghai Banking Telex: 888866 Corporation Limited, Singapore branch; The Hongkong and Shanghai Banking Corporation Limited, Seoul Fax: +44 20 7992 4880 Securities Branch; The Hongkong and Shanghai Banking Corporation Limited, Seoul Branch; HSBC Website: www.research.hsbc.com 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. 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]
35
abc
Global Currency Strategy Research Team

Global
David Bloom Global Head of Currency +44 20 7991 5969 david.bloom@hsbcib.com
Technical Analysis
Murray Gunn +44 20 7991 5384 murray,gunn@hsbcib.com
Precious Metals
James Steel +1 212 525 6515 james.steel@us.hsbc.com
Asia
Richard Yetsenga +852 2996 6565 Perry Kojodjojo +852 2996 6568 Daniel Hui +852 2822 4340 richard.yetsenga@hsbc.com.hk perrykojodjojo@hsbc.com.hk danielpyhui@hsbc.com.hk
United Kingdom
Paul Mackel +44 20 7991 5968 Stacy Williams +44 20 7991 5967 Mark McDonald +44 20 7991 5966 Daniel Fenn +44 20 7991 5003 Mark Austin Consultant paul.mackel@hsbcib.com stacy.williams@hsbcgroup.com mark.mcdonald@hsbcib.com dan.fenn@hsbcib.com
United States
Robert Lynch +1 212 525 3159 Clyde Wardle +1 212 525 3345 robert.lynch@us.hsbc.com clyde.wardle@us.hsbc.com
Marjorie Hernandez +1 212 525 4109 marjorie.hernandez@us.hsbc.com
Main Contributors
Stacy Williams Head of FX Quantitative Strategy HSBC Bank plc +44 20 7991 5967 stacy.williams@hsbcgroup.com 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 responsible for proprietary model trading, concentrating on models based on transactional flow information, high frequency price data and economic activity data.
Daniel Fenn FX Quantitative Strategist HSBC Bank plc +44 20 7991 5003 dan.fenn@hsbcib.com 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.
Mark McDonald FX Quantitative Strategist HSBC Bank plc +44 20 7991 5966 mark.mcdonald@hsbcib.com 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.
David Bloom Global Head of FX Research HSBC Bank plc +44 20 7991 5969 david.bloom@hsbcib.com 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.
Paul Mackel Director of Currency Strategy HSBC Bank plc +44 20 7991 5968 paul.mackel@hsbcib.com 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.

HSBC Risk 282506

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

HSBC Risk 282506

Uploaded by

Copyright:

Available Formats

Quant Special

FX Quantitative Strategy November 2010

Risk on risk off: the full story

Currency FX Quantitative Strategy 10 November 2010

Currency FX Quantitative Strategy 10 November 2010

1. Risk on risk off: the birth of a paradigm

2. What will end the risk on risk off paradigm?

3. Are we risk on or risk off today?

The risk on risk off movie trilogy

Currency FX Quantitative Strategy 10 November 2010

1. Risk on risk off: the birth of a paradigm

Currency FX Quantitative Strategy 10 November 2010

1. The VIX Index since September 2005

Bernanke sub-prime warning

Source: HSBC, Bloomberg

The correlation heat map

Heat map explained

Currency FX Quantitative Strategy 10 November 2010

I. Normal (2005 to mid 2006)

2. I. Normal (2005 to mid 2006)

Source: HSBC, Bloomberg

Currency FX Quantitative Strategy 10 November 2010

Source: HSBC, Bloomberg

Currency FX Quantitative Strategy 10 November 2010

Source: HSBC, Bloomberg

Currency FX Quantitative Strategy 10 November 2010

IV. Attempt to normalize following early crisis pre Lehman

5. IV: Attempt to normalize following early crisis pre Lehmans

Source: HSBC, Bloomberg

Currency FX Quantitative Strategy 10 November 2010

V. Crisis and correlations intensify collapse of Lehman

6. V: Crisis and correlations intensify collapse of Lehman

Source: HSBC, Bloomberg

Currency FX Quantitative Strategy 10 November 2010

VI. Crisis high point strengthening correlations

7. VI. Crisis high point strengthening correlations

Source: HSBC, Bloomberg

Currency FX Quantitative Strategy 10 November 2010

VII. Risk on risk off: high correlations between all markets

8. VII: Risk onrisk off high correlations between all markets

Source: HSBC, Bloomberg

Currency FX Quantitative Strategy 10 November 2010

VIII. Risk on risk off persists

9. VIII: Risk on risk off persists

Source: HSBC, Bloomberg

Currency FX Quantitative Strategy 10 November 2010

2. What will end the risk on risk off paradigm?

Currency FX Quantitative Strategy 10 November 2010

Heat map movie: the directors cut

Source: HSBC, Bloomberg

In a crisis, correlations significantly increase...

Currency FX Quantitative Strategy 10 November 2010

11. Not all crises see correlations rise

Source: HSBC, Bloomberg

Recession and correlation

Currency FX Quantitative Strategy 10 November 2010

Source: HSBC, Bloomberg

Correlation and volatility

Correlations are currently at unprecedented levels

Currency FX Quantitative Strategy 10 November 2010

40 0.3 30 0.2 20 0.1

Source: HSBC, Bloomberg