J.P. 摩根-全球-宏观策略-全球现金与流动性：定位指示器-2019.5.10-188页.pdf

Global Markets Strategy10 May 2019 F&L LibraryPosition indicatorsGlobal Markets StrategyNikolaos Panigirtzoglou AC(44-20)7134-Bloomberg JPMA FLOW J.P.Morgan Securities plcMika Inkinen(44-20)7742 J.P.Morgan Securities plcNishant Poddar,CFA(91-22)6157-J.P.Morgan India Private LimitedSee page 186 for analyst certification and important Below are excerpts of past issues of Flows&Liquidity on the topic of:Position indicators.The excerpts can be found in reverse chronological order,with the most recent update below.The excerpts are not updated and are accurate only as of the date indicated.Generally,we do not email out this document,but maintain it on J.P.Morgan Markets as a reference tool.Duration positions remain skewed to the long side(extract from Flows&Liquidity,03 May 19)As we noted last week in our updated analysis of the global bond supply/demand picture,the shift in the supply/demand balance in 2019 suggested a mildly bearish picture for bonds.Despite this,the yield to worst on the Global Agg index stands nearly 20bp lower than at the start of the year,and touched a low close to a 30bp decline on March 27th.While the steady stream of weaker than expected economic data,as evidenced by the JPM global Economic Activity Surprise Index remaining persistently below zero until mid-April undoubtedly contributed to the decline in yields,we look here at how positioning has contributed.As we noted last week,the retail demand had shifted strongly positive again since the start of 2019.But this inflow in dollar terms does not fully capture the duration impact of those flows.To look at the duration impulse of retail flows,we calculate the duration impact of bond ETF flows by adjusting for the empirical duration of each bond ETF.We use empirical duration rather than actual duration in our calculation,as the latter is either reported vaguely(e.g.by maturity bucket),or not at all.The empirical duration is calculated using the average sensitivity of price returns of each ETF to daily changes in the Bloomberg US Agg index over 1-year rolling periods.We then multiply the flow of each ETF by its empirical duration in years to derive a cumulative bond ETF flow adjusted for duration,and convert the aggregated flows across ETFs to US Agg duration equivalents by dividing the sum by the duration of the Bloomberg US Agg index.When adjusted for empirical duration,shown in Figure 1,the duration impulse from retail investors has been strongly positive YTD and if anything increased modestly over the past few weeks.2Global Markets StrategyFL Library-Position indicators10 May 2019Nikolaos Panigirtzoglou(44-20)7134- Figure 1:Cumulative flows into US bond ETFs in Bloomberg US Agg index duration equivalentsin$bn.Source:Bloomberg,J.P.Morgan.What about institutional investors?Looking at US active bond managers,they also have tended to trade duration on the long side YTD.Figure 2 depicts the beta of the 20 largest active US bond mutual funds relative to their benchmarks,and shows both the 21-day rolling beta that we have typically focused on as well as the 10-day rolling beta that is more responsive when there are sharp market moves.It shows that active US bond mutual funds have tended to exhibit betas above their long-term averages for much of the year,suggesting a long duration bias.The main exception was the immediate aftermath of the March FOMC meeting,when they turned short duration,though they appear to have turned long duration again in mid-April.This picture is broadly consistent with our weekly Treasury Client Survey,where the balance of clients has been net long for much of 2019 to-date,and turned short in before turning long again on April 22nd.Figure 2:21-day and 10-day rolling betas of 20 biggest active US bond mutual fund managers with respect to the US Agg bond indexUnivariate regression of 20 biggest US bond mutual fund managers with respect to the US Agg bond index.Source:Bloomberg,J.P.Morgan.By contrast,European active bond managers appear to have exhibited a short duration bias YTD.Figure 3depicts the beta of the 20 largest active Euro bond mutual funds relative to their benchmarks,and suggests that with the exception of a period from mid-February to early March they have tended to exhibit a short duration bias.The more responsive 10-day rolling beta suggests they briefly turned neutral around end-March,contributing to the rally in global yields at the time,but have since turned short again.-20020406080100120140160Jan-17May-17Sep-17Jan-18May-18Sep-18Jan-19May-190.300.350.400.450.500.550.600.650.700.75Jan-18 Mar-18 May-18 Jul-18Sep-18 Nov-18 Jan-19 Mar-1921-day10-day3Global Markets StrategyFL Library-Position indicators10 May 2019Nikolaos Panigirtzoglou(44-20)7134- Figure 3:21-day and 10-day rolling betas of 20 biggest active Euro bond mutual fund managers with respect to the Euro Agg bond indexUnivariate regression of 20 biggest Euro bond mutual fund managers with respect to the Euro Agg bond index.Source:Bloomberg,J.P.Morgan.What about momentum-based investors as CTAs?Our framework for CTA positioning in Tables A5 and A6 in the Appendix suggests they have traded duration on the long side for much of the year.Figure 4 shows that the z-scores at the front-end of the US,have gradually drifted higher until late March.Similarly,Figure 5 depicts z-scores for the 10y UST and Bund futures contracts,and shows that after briefly turning short in end-February/early March,CTAs likely contributed to the March rally by re-entering duration longs.Our momentum signals for German bond futures also suggest CTAs have been long duration for much of the year.More recently,short term momentum in 10y USTs turned negative on Thursday,suggesting a risk that CTAs could start to reduce long duration exposure.The betas of risk parity funds to bond returns also suggest long duration exposures for much of the year,though since the start of April their exposures have turned more neutral(Figure 6).Figure 4:Momentum signals for 2y and 5y USTsz-score of the momentum signal in our Trend Following Strategy framework shown in Tables A5 and A6 in the Appendix.Solid lines are for the shorter term and dotted lines for longer-term momentum.Source:Bloomberg,J.P.Morgan.Figure 5:Momentum signals for 10y USTs and Bundsz-score of the momentum signal in our Trend Following Strategy framework shown in Tables A5 and A6 in the Appendix.Solid lines are for the shorter term and dotted lines for longer-term momentum.Source:Bloomberg,J.P.Morgan.-0.100.000.100.200.300.400.50Jan-18 Mar-18 May-18 Jul-18Sep-18 Nov-18 Jan-19 Mar-1921-day10-day-1.5-1.0-0.50.00.51.01.5Jan-18 Mar-18 May-18Jul-18Sep-18 Nov-18 Jan-19 Mar-19 May-192Y USTs5Y USTs-1.5-1.0-0.50.00.51.01.52.0Jul-18Sep-18Nov-18Jan-19Mar-19May-1910Y USTs10y Bunds4Global Markets StrategyFL Library-Position indicators10 May 2019Nikolaos Panigirtzoglou(44-20)7134- Figure 6:Bond beta of risky parity fundsRolling 21-day bond betas based on a bivariate regression of the daily returns of our Risk Parity fund return index to the daily returns of the S&P 500 and Bloomberg/Barcap US Agg indices.Given that these funds invest in both equities and bonds we believe that the bivariate regression will be more suitable for these funds.Our risk parity index consists of 25 daily reporting Risk Parity funds.Source:Bloomberg,J.P.Morgan.In all,our position indicators suggest long duration positions among retail investors,active US bond funds,momentum-based investors such as CTAs and risk parity funds have all contributed to the bond market rally this year despite the modest deterioration in the supply/demand picture.The main exception has been Euro active managers,who appear to have traded duration from the short side for much of 2019.How extreme are European underweights post the ECB?(extract from Flows&Liquidity,08 Mar 19)In the equity space,hedge funds appear to have increased their negative stance on European banks over the past week,shifting towards more extreme negative territory.ETF investors have become progressively more underweight on European equities.Their current underweight stands at more than a decade low.In the euro,CFTC data suggest that heavy spec shorts vs.the dollar are not only confined to hedge funds but extend to non-hedge fund investors also.This weeks bond rally may have been exacerbated by short covering by active Euro bond funds and a shift back towards long duration positions in USTs by CTAs.More confirmation from US Flow of Funds that the 2018 US repatriation episode is largely behind us.Our best guess is that last years US repatriation flow of$430bn was mostly used for share buybacks,by around 2/3rds,with the rest split between corporate bond withdrawal and capex.As the US repatriation flow appears to be largely behind us,this raises the prospect of downside risk to US share buybacks and capex growth,and upside risk to US corporate bond issuance,relative to last year.TLTRO 3s provide a meaningful extension of existinglow cost funding,and creates around 120bn of additional borrowing space for periphery banks on aggregate.The market reaction to this weeks ECB meeting was rather puzzling.The collapse in European bank stocks in particular was in our mind at odds with the ECBs generosity which effectively allows banks to secure cheap long-term funding until 2023.This market reaction seemed even more puzzling for Italian banks given the rally in Italian government bonds.What types of investors were likely responsible for this market reaction?Looking at EMEA sector positioning from J.P.Morgans Prime Finance clients,it looks like hedge funds were the most likely culprits.Figure 7shows that although hedge funds clients were rather underweight on European financials ahead of the ECB meeting,they increased their underweight position by even more post the ECB meeting.As a result their underweight position has become more extreme at 1.5x stdevs currently.Figure 7:L/S ratio by sector in EMEA based on J.P.Morgan PrimeFinance client positioning z-score;see page on J.P.Morgan Prime Finance client positioning in the Appendix for descriptionSource:J.P.Morgan.The extremity of investors underweights on European financials is also reflected in the share of Europeanbanks in equity indices.This is shown in Figure 2 which depicts the difference in the share of banks in the-1.5-1.0-0.50.00.51.01.52.02.5Jan-18Apr-18Jul-18Oct-18Jan-19Apr-19Risk Parity beta to bonds-2.0-1.00.01.02.03.0FinancialsUtilitiesReal EstateIndustrialsInfo TechCons Disc.Telecom SrvcsEnergyHealthcareMaterialsCons Staples27-Feb6-Mar5Global Markets StrategyFL Library-Position indicators10 May 2019Nikolaos Panigirtzoglou(44-20)7134- Stoxx600 index minus the share of banks in the MSCI AC World index.This difference has breached previous historical lows seen during the Euro debt crisis in the summer of 2012 and ahead of the French presidential election at the beginning of 2017.What about real money or retail investors stance?To gauge real money or retail investors stance we typically look at ETF based position indicators.However there are no ETFs on European banks that are large enough to be reliable indicators in our mind.By looking at the broader universe of European equity ETFs we find that ETF investors have become progressively more underweight on European equities.This is shown in Figure 9 which shows the European share in equity ETFs divided by the European share in global equity indices.This equity position proxy has been declining steeply during both 2018 and 2019 to a level that is currently well below the low previously seen during the Euro debt crisis.What about investors positioning on the euro?Figure 4 shows two types of spec positions on EURUSD futures based on CFTC data:the spec positions for the Non-Commercial category(i.e.the overall universe of specinvestors)and the spec positions for Leveraged Funds(i.e.hedge funds).The two indicators have been moving in similar direction over the past year or so but hedge funds appear to have been generally more negative on the euro and earlier than non-hedge fund investors.Hedge funds short base on the euro appears to have been little changed since last October at close to 20%of open interest.It was rather non-hedge fund investors that appear to have become progressively more negative on the euro over the past few months including this past week.The negative of non-hedge fund investors likely converged to that of hedge funds after this weeks ECB meeting.Figure 8:Share of banks in Stoxx600 over the share of banks in the MSCI AC World index”Monthly observations.Mar19 observation as of 7th March 2019.Source:Bloomberg,DatastreamFigure 9:European share in equity ETFs over European share in the MSCI AC World indexMonthly observations.Mar19 observation as of 7th March 2019.Source:Bloomberg,Datastream1.01.11.21.31.41.51.61.71.81011121314151617181930%32%34%36%38%40%42%44%46%050607080910111213141516171819European equity ETF share over European share in MSCI AC world equity indices 6Global Markets StrategyFL Library-Position indicators10 May 2019Nikolaos Panigirtzoglou(44-20)7134- Figure 10:Spec positions on EURUSD futures EURUSD spec positions for Non-Commercial(i.e.overall spec)and Leveraged Funds(i.e.hedge funds)categories as reported by CFTC.Last CFTC reported obs is for 26th Feb 2019.The dotted lines are extrapolated values for 5th March and 7th March respectively based on changes in the open interest.Source:CFTCWhat about investors positioning in Euro bonds?The past week has seen a marked rally in bond markets into the ECB meeting.We had flagged in recent weeks that positioning in Bunds looked elevated,making this weeks reaction somewhat harder to reconcile.We had noted over the previous three weeks that our regular metrics for momentum-based investor positioning(Exhibits A5 and A6)have flagged quite extended duration longs in 10y Bunds and somewhat more neutral positions on 10y USTs.Indeed,we noted last week that there were signs these extended Bund longs had begun to trigger mean reversion signals,and that UST momentum was starting to turn bearish.This weeks rally has reversed both of those shifts,with Bund momentum again rising close to extreme levels(Figure 11).For UST futures,both the shorter-and longer-term signals turned short last week,only to switch again this week suggesting that position shifts by CTAs might have exacerbated volatility in bond markets over the past two weeks.Figure 11:Momentum signals for 10Y UST and Bundsz-score of the momentum signal in our Trend Following Strategy framework shown in Tables A5 and A6 in the Appendix.Solid lines are for the shorter term and dotted lines for longer-term momentum.Source:Bloomberg,J.P.MorganWhat about real money active bond managers?Both Euro and US active bond managers had shifted from a more bearish duration bias in early February to quite elevated betas towards the end of February.This is shown in Figure 12 and Figure 13,which depicts the beta of the 20 largest active Eu