MARKET RISK REPORT BACK TO NORMAL?

Size: px

Start display at page:

Download "MARKET RISK REPORT BACK TO NORMAL?"

Russell Doyle
5 years ago
Views:

1 March 2013 MARKET RISK REPORT BACK TO NORMAL? Executive Summary February started positively but quickly saw corrections as the Italian election produced an inconclusive result, the US sequestration caused concerns and European PMI data was weak. However, QE, higher risk appetite and a general asset rotation continued to push cash out of low-yielding bonds and into risky assets including equities. Asset-specific issues were again more important than systemic (e.g. risk-on/risk-off). Inter-market correlation rose slightly but remained near 12-month lows. Realised volatility (of Eurostoxx equities over 30 days) rose from below 7% at the start of the year (which we had put down to unusual seasonal effects) to 12% at the end of January to 23% at the end of last month. Volatility once again rose in almost all asset classes. Equity markets rose again, but at a much lower pace than the previous month. Volatilities rose across the board: to 15% in Europe, and 12.6% in the global Financials sector. Sovereign bond prices diverged again during the month with Italians falling and Germans rising post the Italian elections. German bond volatility fell to 5.3% but Italian rose to 15.2%. In FX, the big mover this month was the Euro which fell sharply on the Italian elections and weak PMI data. Volatility fell slightly though with $/ down to 8.3%. Option volatility will have been higher in general due to the large spike in the volatility of implied volatility. Underlying asset price moves were also quite large, with the exception of some equity markets. Commodities prices dropped again on weaker macro news and light but generally positive geo-political news. Volatilities rose as prices fell: Oil volatility rose from 10.4% to 14.0%. Real Estate (share) prices paused their earlier rally (with the exception of Japan where prices rose a massive 12% on the month) and volatilities rose across the board: the US to 11% and Japan rose to 28%. Special Feature: The impact of correlation on risk within equity portfolios Author: Niall O Connor Page 1

2 Summary volatility matrix Almost every asset class saw a collapse in volatility during December and almost every asset class fell into the lower third of its 12- month volatility range. We attributed this to traders sitting on hands ahead of the US fiscal cliff and a certain amount of lack of energy and profit-taking. The large drop in volatilities does appear to have been a seasonal issue, and volatilities rose for most asset classes during both of the last two months. Our urging of caution last month proved prophetic. We would re-iterate what we had said last month: there is still a lot of work to be done to avert big recessions in Europe, the UK and Japan. Budget deficits remain very large in the US, UK, Japan and Spain. QE/financial repression is giving rise to yield-chasing bubbles (and remember what happened in 2007?) in corporate bonds, property and possibly even in equities. Currencies including the Swiss Franc and Aussie Dollar remain seriously overvalued, distorting their local economies. Shorter term things had started to look better but the Italian election result and weak macro data have given markets pause for thought. Low volatility would still look odd to us in the context of so much uncertainty. ASSET CLASS AREA LATEST VOLATILITY LATEST, Z- Z SCORE REGIME EQUITIES North America 11.8% -0.1 medium Far East 19.4% 2.1 HIGH Europe 15.0% 0.2 medium Japan 23.1% 2.5 HIGH Energy 14.3% -0.4 low Consumer Staples 8.9% -0.2 medium Financials 12.6% -0.6 low IT 11.1% -1.1 low VOLATILITY OF Volatility of VIX 166% 3.7 HIGH VOLATILITY Volatility of VSTOXX 146% 2.9 HIGH GOVERNMENT BONDS Germany 5.3% -0.8 low US 2.9% -1.4 low Japan 1.9% 0.5 HIGH Italy (rh axis) 15.2% 0.8 HIGH FX VS $ $/ 8.3% 0.1 medium Yen/ 17.5% 2.3 HIGH CHF/$ 7.0% -0.5 low UK / 12.2% 3.5 HIGH COMMODITIES Oil (Brent) 14.0% -1.1 low Gold 14.1% -0.3 medium Copper 14.8% -1.1 low PROPERTY US (GPR250) 10.9% -0.3 medium Germany (GPR250) 16.0% -0.5 low Japan (GPR250) 28.2% 2.0 HIGH ALTERNATIVES HFRX Global HF (EUR) 2.5% 0.6 HIGH Avg PE Fund 9.0% 0.3 medium KURTOSIS ZCF 1% left (vs for normal curve) low CORRELATION Average market correlation with euro equities 43% -0.9 low Note: Throughout the text we refer to volatilities as being "low", "medium" or "high". We define this by defining three equal "sized" regimes over the last 12 months. i.e. high volatility implies that volatility is in the upper third of its statistical range over the last 12 months. The table shows the "z-score" of the volatility of each market, i.e. how many standard deviations above (or below) the mean over the last 12 months each market's volatility is. Page 2

3 Key News (Major Volatility-Driving Events) The elections in Italy led to a hung parliament: markets wobbled and Italian stocks fell and bond yields rose sharply. The US sequestration (what we had labelled as fiscal cliff 2 ) also caused concerns. Equity markets were supported by the continuing re-appearance of mega-deals, no doubt aided by record low corporate borrowing rates. Berkshire Hathaway bid $28bn for Heinz, the largest ever food industry deal, and Comcast bought out the minorities in its NBC Universal subsidiary for $16.7bn and CNOOC was given the go-ahead to acquire Nexen for $15.1bn. These came after Dell had announced plans to take itself private for $24.4bn and Liberty Global agreed to buy Virgin Media for $23.3bn. American Airlines and US Airways announced a merger, as did Office Depot and OfficeMax. Currency markets remained volatile due to a confusing statement issued by the G7. The UK s Bank of England minutes showed a push for more QE: the stock markets and the pound fell. Weak UK manufacturing data also put downward pressure on sterling. Gold continued to lose its shine as investors move away from haven and safe assets. The price dropped below $1,600. GDP fell by a disappointing -0.6% q/q in the euro area during Q Japanese GDP also fell, by -0.4% annualised, the third quarter of decline in a row. European PMI s continued their weak trends: suggesting -0.3% GDP growth for the Eurozone as a whole. Germany remained at expansionary levels, France s was surprisingly weak and remains in contraction. The Chinese PMI once again disappointed: dropping to 50.4 for February, although it does remain in expansionary territory (just) and the Chinese New Year is probably at least part to blame. GDP forecasts for 2013remained largely unchanged on the month. 3% 2% 1% 8% 6% 4% 2% US Japan Euro Area Germany France 0% 0% Dec-11 Mar-12 May-12 Jul-12 Aug-12 Oct-12 Dec-12 Mar-13 May-13-2% -1% -4% -2% -6% Italy Spain UK China (rh axis) Page 3

Kurtosis and correlation in the equity markets The large deviations from normal in the distribution of returns in the equity markets we have seen over the last two months all but disappeared by the

4 Kurtosis and correlation in the equity markets The large deviations from normal in the distribution of returns in the equity markets we have seen over the last two months all but disappeared by the end of February. We had said that the very low levels of volatility at the end of last year and at the start of 2013 had probably made the big deviations less meaningful than they could have been. Now volatility levels are at less extremely low levels we will begin to take more notice of non-normality. Although at the present time markets are very close to normal ZCF 1% left Z 1% left ZCF 1% right Z 1% right Methodology To capture a measure of both Skewness and Kurtosis we look at Cornish-Fischer expansion, which gives a good measure of the tails of the equity market. (We use a 60-calendar day rolling basis). The underlying market we plot is the Eurostoxx 50, but other European equity markets are likely to show very similar results. We plot on the chart the expected Z-scores for 1% left tail (i.e. a 99% VaR) and a 1% right tail assuming a normal distribution: +/ We also show the Cornish-Fischer expansion result for the same market. This indicates how far from a normal distribution each tail was. On a long term basis on average the tails are slightly fatter than the normal distribution would suggest, which should not come as a surprise. What is perhaps more surprising is how much variation in fat-tailedness there has been: a 99% VaR has varied between -1 and -4.5 standard deviations over time. The biggest variations from +/-2.33 came in 2008 and Inter-market correlations Correlation (of asset classes with European equities) was unusually high from June to November 2011 as a result of the risk-on/risk-off pattern of trading: just about all markets were correlated to this pattern. Correlation fell significantly as markets normalised at the end of 2012, dragged down by very low currency correlation with equity markets. During the month correlations rose slightly driven by increasing correlation between equities and /$. It still remains at low levels relative to the average of the last year. Note: The chart shows 30-day correlation over time between different markets and the European equity market. Higher levels of correlation will in general lead to less ability to diversify risks, and higher portfolio volatility for given position holding volatility. Page 4

The impact of correlation on risk within equity portfolios Last year we looked at how correlation between equities in a portfolio impacted the overall volatility of a portfolio.

portfolio's volatility. Most risk models focus on the likely volatility of individual stocks, but assume the correlation matrix remains constant.

5 The impact of correlation on risk within equity portfolios Last year we looked at how correlation between equities in a portfolio impacted the overall volatility of a portfolio. And in particular we looked at how diversification reduced the volatility of a portfolio relative to its constituents and saw that isolating diversification shows it can have considerable impact on a portfolio's volatility. Most risk models focus on the likely volatility of individual stocks, but assume the correlation matrix remains constant. The problem with this approach is that during a crisis two things occur simultaneously both of which serve to increase portfolio volatility; namely individual stock volatility rises, but also, and perhaps as significantly, diversification effects significantly diminish: the "all correlations go to 1" during a crisis effect. To show the actual impact we look at a sample large-cap equity portfolio. The portfolio we have chosen is an arbitrary one, consisting of an equal weighting of 10 Eurostoxx 50 stocks, from a representative range of industries and countries. We would expect our portfolio to mimic the Eurostoxx 50 (and other broader large-cap indices) relatively well in terms of both volatility and performance. At the end of January our chosen basket had a 30d historic volatility of 14%, compared to the average volatility of the constituents of 22%, with a range of 13% for Unilever to 32% for Carrefour. We define the diversification effect to be the ratio of the volatility of our portfolio divided by the average volatility of individual stocks: effectively this is also a measure of correlation between them. The lower this figure the lower the correlation and the higher the benefits to the portfolio of diversification are. The chart on the left shows these three sets of data over the last two years. Correlation between the stocks was very high for most of 2011 during the height of the European crisis and portfolio volatility was close to the average of the stock volatilities. Since the start of this year the risk-on/risk-off trade has been replaced by a more stock-specific set of returns (driven in part by the ending of the crisis but also as the amount of stock-specific news has been much higher). Correlation has dropped and portfolio volatility has dropped to just over 60% of the average stock. We see a surprisingly similar pattern in a similarly-constructed US equity portfolio with high correlation for most of 2011 and a substantial drop in Page 5

6 The impact of correlation on risk within equity portfolios It is also instructive to look at how correlation has varied as a function of average stock volatility (here for Europe again) Diverication effect % 10% 20% 30% 40% 50% 60% 70% Average stock volatility Over the last two years higher volatility regimes have coincided with higher correlation (again the all correlations go to 1 in a crisis effect). In one way the change in correlation does not matter as we tend to think in terms of portfolio or index volatility, and not stock volatility. Reversing our previous argument will say that individual stock volatility varies less than portfolio volatility as correlation is itself correlated with individual stock volatility. i.e. portfolio volatility= stock volatility plus increase in correlation becomes stock volatility= portfolio volatility less increase in correlation. But we think it is important not to forget the fact that when crises hit not only do stock volatilities rise but correlation does too, so portfolio volatility rises disproportionately. Almost all risk models assume constant correlation matrices. And it is also difficult to adequately define stress tests to incorporate changes to the correlation matrix without simply resorting to everything being +/-1. This method tends to give disconcertingly high volatility, although in practice stress test results using this method might not have been far off what subsequently happened during the financial crisis. It is hard enough to predict portfolio volatility during normal times, but harder still in times of genuine stress. Remembering that correlations rise as volatility does will help. Page 6

Equities Prices rose in the first half of the month driven by a general rotation away from safe assets towards more risky and higher-yielding assets.

Prices: Sectors Price changes were mixed by sector but in general most sectors were close to flat on the month with the gains in the first half being lost in the second.

Volatility also rose in all other regions.

7 Equities Prices rose in the first half of the month driven by a general rotation away from safe assets towards more risky and higher-yielding assets. During the second half of the month the inconclusive result of the Italian elections and weak European macro data caused markets to fall back somewhat. Prices: Sectors Price changes were mixed by sector but in general most sectors were close to flat on the month with the gains in the first half being lost in the second. Volatility: Regions With the markets refocused on the European crisis after the Italian elections unsurprisingly it was volatility in Europe which saw the biggest rise on the month. Volatility also rose in all other regions. Volatility rose from low to medium in North America and Europe, and it remained high in the Far East and Japan in comparison with the last 12 months' average. Volatility: Sectors On a sectoral basis volatility rose in almost all sectors on the month. The exception was the IT sector, where previously high newsflow abated more recently. Volatilities remain medium to low on a 12-month basis in our chosen sectors. Page 7

Equities 40 Implied Volatility (Market-Implied Near Term Outlook) Implied Volatility, % 30 20 10 VSTOXX VIX Implied Volatilities spiked during the month before falling back somewhat: to 15.

8 Equities 40 Implied Volatility (Market-Implied Near Term Outlook) Implied Volatility, % VSTOXX VIX Implied Volatilities spiked during the month before falling back somewhat: to 15.5% for the US (VIX), and 21.0% for Europe (VSTOXX). - The return in realised volatility to higher, more normal levels exceeded the rise in implied vol and the implied/realised ratio (for Europe) fell back below 100%. At the end of the month it was at 92%, suggesting that the market now sees less volatility-inducing news over the next 30 days. Page 8

Fixed Income (10 Year Government Bonds) Yield 8 7 6 5 Spain yield Prices European fixed income markets remained calm until the Italian elections, at which point Italian yields rose and German yields

4 US and Japanese bond prices rose on the month. Volatility The ECB's promise of intervention (if required) had made peripheral bonds trade range-bound, and hence volatility dropped.

9 Fixed Income (10 Year Government Bonds) Yield Spain yield Prices European fixed income markets remained calm until the Italian elections, at which point Italian yields rose and German yields fell very sharply. We took a step back towards the old European crisis risk-on/risk-off trade, although Spanish yields were less affected than previously. 4 US and Japanese bond prices rose on the month. Volatility The ECB's promise of intervention (if required) had made peripheral bonds trade range-bound, and hence volatility dropped. The Italian election result (and weak European macro data this month) probably reminded people that the European crisis is not yet over: low/no growth and indecisive or anti-austerity governments could still cause problems. Volatilities are now mixed: Germany (5.3%; low), Italy (15.2%, high), the US (2.9%; low) and Japan (1.9%; high). Page 9

FX Prices The G7 s statement confused more than it clarified by warning against currency wars whilst simultaneously seeming to support Japan s weakening policies.

10 FX Prices The G7 s statement confused more than it clarified by warning against currency wars whilst simultaneously seeming to support Japan s weakening policies. Despite this, FX volatility actually fell slightly during the month as the Yen changed direction and strengthened slightly, at least relative to a falling Euro which itself was driven down by the Italian election result and more weak macro data points. Volatility Volatility in $/ fell to 8.3% (medium), Yen/ to 17.5% (still high), and / rose sharply, nearly doubling to 12.2% (high). We had hypothesised that currency wars had reached an apparent stalemate and also interest rate differentials are very low and hence definitionally stable too. However, the reality of the situation has proved very different over the last two months, particularly with strong intervention in Japan and a shift towards a higher inflation target driving the Yen down. The Italian election result also shook off some of the earlier complacency that the European crisis is over. Note: The charts show currencies vs. the. Axes on the first chart are inverted to show conventional currency quotations, but with higher on the chart representing a stronger currency vs. the euro. Page 10

(Equity) Options Annualised Volatility (1-month average) 180% 160% 140% 120% 100% 80% 60% 40% 20% 0% Volatility of VIX Volatility of VSTOXX Volatility of Implied Volatility The spike and subsequent

11 (Equity) Options Annualised Volatility (1-month average) 180% 160% 140% 120% 100% 80% 60% 40% 20% 0% Volatility of VIX Volatility of VSTOXX Volatility of Implied Volatility The spike and subsequent drop in implied volatility during the month caused both US and European volatility of implied volatility to rise very sharply on the month: both more than doubling. VIX volatility rose to 166% and VSTOXX to 146% at the end of the month. Both are now high relative to their 12-month averages. Major (Regional Equity) price moves Option volatility is likely to have risen during the month as volatility of volatility followed that pattern. Directional price moves in equities were relatively muted during the month after a period during which they had been rising. Options on bonds and FX would have probably seen big changes in delta due to the big moves in most of those underlying markets. Note on Treatment: Options show more complex behaviour than the other instruments we look at in this report, so we make some simplifying assumptions. As Calls and Puts are in effect polar opposites and in and out of the money options behave very differently, it is hard to generalise all options behaviour. However, we look at the two key drivers: volatility of implied volatility and major price movements of the underlying security. Implied volatility (via an option's Vega) drives option prices, so a big indicator of option price volatility is the volatility of implied volatility. But usually the biggest driver of individual option prices is the movement of the underlying (via the option Delta): a move in either direction will cause the option to go more in or out of the money (and a corresponding change in the option s Delta and price volatility). As a proxy for this, we look at the 30-day price swing of equity market indices; options on bonds or FX could of course behave differently. The 30-day period is relatively close to the time to maturity of many options. Calls and Puts will respond in opposite fashions: calls becoming more volatile (relative to the size of the underlying notional) as prices rise. Note on Convertibles: Convertibles are in effect a combination of a bond and a call option, with the bond portion usually making little contribution to the instrument volatility unless the option is significantly out of the money. As such, convertible portfolios volatilities will tend to behave similarly to call option portfolios, and this commentary can be applied to convertibles as well as options. Page 11

Commodities Price (Near Future) 125 115 105 95 85 2,000 1,900 1,800 1,700 1,600 1,500 1,400 1,300 Oil (Brent) Gold (rh axis) Prices Most commodities prices dropped sharply during the month as

12 Commodities Price (Near Future) ,000 1,900 1,800 1,700 1,600 1,500 1,400 1,300 Oil (Brent) Gold (rh axis) Prices Most commodities prices dropped sharply during the month as economic news came in relatively weak, and investors seemed to continue to be happy to move into riskier asset classes and away from safe havens, in particular Gold. 75 1,200 1, ,000 Volatility The lack of any real geopolitical news during the month kept volatility lower, although this was offset in the second half of the month by the drop in asset prices. Volatilities rose over the month. Oil volatility rose to 14.0% (low). Gold rose marginally to 14.1% (now medium) and Copper rose slightly to 14.8% (low). Page 12

Real Estate (Real Estate Share Prices) and Alternatives Prices Real Estate equity markets were generally flat on the month with the exception of Japan, where they rose a substantial 12%, and this on

13 Real Estate (Real Estate Share Prices) and Alternatives Prices Real Estate equity markets were generally flat on the month with the exception of Japan, where they rose a substantial 12%, and this on top of an already impressive rally year to date. Volatility Volatilities rose in all major markets: US volatility rose to 10.9% (now medium), Germany rose to 16.0% (still low) while Japan rose to 28.2% (still high). Note: Note that for property we look at indices of the share prices of property companies, and not the underlying property directly, for which little real-time data is available. This is usually consistent with funds which tend to invest in property indirectly, e.g. via REIT s or property companies. As REIT s are usually focussed on commercial property, residential property may also follow a slightly different pattern to that discussed in this article. Price Index ,100 1,090 1,080 1,070 1,060 1,050 1,040 Powershares Global Listed PE Fund HFRX Global HF (EUR) Other Alternative Investments The rally in Alternatives continued, albeit not as strongly as the previous month, as cash continued to chase yield ,030 1, ,010 14% Annualised Volatility (1-month average) 12% 10% 8% 6% 4% 2% HFRX Global HF (EUR) Avg PE Fund Volatility levels in both the global hedge fund indices (HFRX) and Private Equity (PE) funds rose on the month: HFRX volatility to 2.5% (now high) and an average of PE funds rose to 9.0% (medium by the standards of the last 12 months). 0% Page 13

14 NOTES Definitions To avoid repetitions, the term volatility refers to annualised, 30-day average realised volatility in local currency unless otherwise specified. As such, it may be lower than, and lag, shorterterm market volatility in times of high market volatility. Charts show data up until 31st January 2013, and the commentary was written on or before 12th January Disclaimer The commentary does not constitute, and is not intended to constitute, investment advice. Any views expressed in this report are based on historical market data and as such cannot be interpreted as being forward-looking, or to constitute forecasts. Past movements are not necessarily indicative of future movements. Employees of IRML may hold positions in securities mentioned. All expressions of opinion reflect the judgment of IRML at this date and are subject to change. Information has been obtained from sources considered reliable, but we do not guarantee that the report is accurate or complete. This document is not for US clients or distribution to the US. IRML This document is the property of IRML and should not be copied or distributed to any third party without the prior consent of IRML please contact us re distribution rights.

MARKET RISK REPORT Creeping Upwards May 2018

MARKET RISK REPORT Creeping Upwards May 2018 Executive summary Volatilities in general declined again their very elevated levels of the prior months. Asset volatilities were mixed relative to their 12-month