Drivers of the EUR long end

Transcription

1 Austria Belgium Germany Spain Finland France Netherlan ds Total assets (EUR, bns) 2 July 212 Fixed Income Research Drivers of the EUR long end European Interest Rate Strategy Research Analysts Thushka Maharaj thushka.maharaj@credit-suisse.com Florian Weber florian.weber@credit-suisse.com Regime shift to steeper long end There are many opposing forces driving the slope of the EUR long end. We expect regulatory reform to be a key driver going forward. We summarise the main macro drivers in the current sovereign crisis such as pension fund ALM hedging, CVA demand, and the changing regulatory landscape. If the proposed changes to the liability discount curve are implemented, we expect a 6%-1% improvement in Dutch pension fund coverage ratios. Normalisation in rates, smaller exotic dealer flows, reduced need for CVA receiving, and the shift in regulation towards higher discount curves support steepening of the EUR 1s3s and 2s3s curves both in swaps and cash. Purely based on the new proposed discount curve, the EUR 1s3s curve should be 4bp steeper. We recommend several trades related to this curve-steepening theme. Exhibit 1: Insurance companies and pension fund assets are significant relative to GDP Numbers above show % of country GDP Exhibit 2: Proposed change in the discount curve likely to reduce demand for the ultra-long end 25 Pension fund total assets 3.5% Insurance corporations % 2.5% 2.% 1.5% %.5%.% 2-3 Spot rate 2-6 Spot rate Current spot curve Maturity Source: ECB, Credit Suisse Source: the BLOOMBERG PROFESSIONAL service, Credit Suisse ANALYST CERTIFICATIONS AND IMPORTANT DISCLOSURES ARE IN THE DISCLOSURE APPENDIX. FOR OTHER IMPORTANT DISCLOSURES, PLEASE REFER TO CREDIT SUISSE SECURITIES RESEARCH & ANALYTICS BEYOND INFORMATION Client-Driven Solutions, Insights, and Access

2 Table of Contents The EUR long end and the race to financial repression 3 Trade recommendations 2y is the new 3y... 4 Main players 6 Drivers of the EUR long end 11 Drivers during the financial crisis 15 Exotic gamma hedging and the gamma trap Pension fund asset-liability management Drivers during the sovereign crisis 17 CVA hedging flows Regulatory changes in the pension and insurance sector are key 19 Solvency II demand for the long end shifting to 2y point Pension fund reform converging to Solvency II Market implications of regulatory changes 2y the new sweet spot Drivers of the EUR long end 2

3 The EUR long end and the race to financial repression We review the main drivers of the EUR long end and how these drivers have evolved from the financial crisis to the current sovereign crisis. We also discuss how the low rate environment and the imposition of financial repression has negatively affected coverage ratios for pension funds. Both low rates and extreme volatility have forced regulatory changes that we expect to have far reaching consequences for fixed income markets. With assets of over 4 trillion within the pensions and insurance sector, we expect pension fund reform to be a key theme for the EUR long end. These developments have brought long-end dynamics sharply into focus once again. It has now become normal for 3y rates to move 5bp-1bp a day. The 1s3s curve has been equally volatile, with a 5bp flattening or steepening not abnormal anymore. In this report we look at the evolution of the drivers of the EUR long end from the pre-crisis dynamics to the financial crisis and now during the current sovereign crisis. We also discuss recent regulatory changes and their expected impact on demand for the long end. We estimate that pension fund coverage ratios improve by up to 1% on the proposed change in the discount curve. This could translate into a 5-6bn reduction in hedging demand in the long end. We delve into the following topics in this report: The Gamma Trap : We discuss the main exotic positions held before the crisis and how these positions led to inversion of the 1s3s curve during the financial crisis. Sovereigns no longer risk free and CVA hedging flows: We also discuss the emergence of counterparty credit risk following the collapse of Lehman Brothers in 28. With the recent volatility in sovereign markets, we have had to come to terms with sovereigns having meaningful counterparty credit risk. We discuss how counterparty valuation adjustment (CVA) hedging impacts the EUR long end. Expected changes towards two-way CSA agreements between banks and sovereigns should help reduce banks counterparty credit risk in transactions with governments. Demand from CVA desks has already fallen, and we expect less need for CVA hedging going forward. Low rates forcing pension fund reform away from marking to market: We discuss pension fund links with the long end and discuss recent regulatory changes. Regulators in Europe realised that the low rate environment put their pension plan returns at risk and have proposed wide-sweeping reforms. We summarise the main measures proposed for insurance companies under Solvency II, the Dutch pension fund (FTK II) proposal, and various measures announced in Sweden, Denmark, and Finland. The Danish announcements were particularly important for the market as Danish pension funds are known to be better funded, on average, than Dutch funds. So despite the fact that Danish funds are smaller than Dutch pension plans, the disproportionate need to unwind hedges in Denmark has caused sharp bearish steepening of the EUR swap curve. Confluence of factors support further EUR 1s3s steepening Solvency II regulation for insurance sector: the current proposal is to use the EUR swap curve out to 2y and thereafter a modeled rate that ultimately converges to the ultimate forward rate (UFR) of 4.2%. Thus demand for fixed income past the 2y point is likely to fall. Pension fund regulatory reform is moving in the direction of using the swap curve out to 2y and thereafter a modeled rate to discount liabilities. The main takeaway from these recent proposals is faster convergence to the Solvency II framework and a change in the discounting curve used to value liabilities. All the measures act to reduce the value of liabilities and hence the need to hedge using long-dated fixed income instruments. We expect this to reduce demand for the ultra-long end. Drivers of the EUR long end 3

4 Gradual move to defined contribution instead of defined benefit schemes more flexibility with risk taking and less need to provide guaranteed returns reduce the need to hedge using fixed income assets. Reduced CVA hedging demand: banks have reduced positioning with sovereigns since the start of the 28 crisis. This is expected to reduce the receiving pressure in the EUR long end. Reduced hedging need by exotic gamma desks. Normalisation of curve as crisis fades. This has been a structural theme we discussed in our 212 Outlook. Trade recommendations 2y is the new 3y Taking all of the above into account, our view is that long-dated forwards will likely grind higher. Potential regulatory changes, reduced CVA hedging needs, and reflation risks all point towards higher long-dated forwards. Here we summarise the main trading themes and in section Market Implications we detail our specific trade ideas. Outright level of yields We expect the 2y point to benefit the most from the proposed change in regulations. We recommend paying EUR 2y2y outright (current mid is 2.34%) as a simple expression of the above view. These long-dated forwards are at historical lows and significantly below levels predicted by the newly proposed discount curves that we discuss in more detail later. Curve/curvature We expect the 2s3s curve to steepen as insurers gradually move their hedges out of 3y into the 2y sector. We expect the EUR 1s2s3s fly to richen, the EUR 2y1y rate to rise, and longdated forwards to move higher. We recommended paying EUR 1s3s, 2y forward as part of our 212 Outlook. The curve has steepened 3bp since then. Although entry levels may not look attractive, we still hold this as a structural view. We recommend shortening duration in both the Dutch and German curves. We find the switch into the DBR 34s from the DBR 42s cash-for-cash an efficient expression of the view. We also expect the cash 2s3s curve to steepen in both the Netherlands and Germany. Volatility We also expect vol on EUR 2y tails to richen relative to 3y tails. Skew on 2y tails is likely to be biased towards receivers. Note that in a grind higher environment where receiving needs are reduced, the need to buy receivers is also likely reduced. This suggests that the bottom right part of the vega surface can grind lower. Selling ATM receivers on 1y3y or similar options. This idea expresses the view that we have seen the lows in long-dated rates, but on the other hand they are unlikely to move significantly higher; therefore, selling a receiver is preferable to buying a payer, for example. Drivers of the EUR long end 4

5 Selling 4.2% payers outright. With vol being reasonably high, investors can get a decent premium by selling 4.2% payers outright. This can be used as a hedge against existing steepening positions (or outright bearish ones). We choose the 4.2% strike on purpose given that this is the theoretical discount rate proposed for liabilities longer than 2y. There is therefore a reasonable chance that long forwards gravitate towards that number. Buy payer flies. We are currently recommending a 9m3y payer fly (see EST 2-Feb-12 and EST 22-June-12 for more details). Cross market Receive 2y2y GBP versus EUR. We do not expect GBP forwards to move much, while EUR can reprice higher. We think there is plenty of room for the EUR-GBP 2y2y spread to rise. Drivers of the EUR long end 5

6 Main players The main players in the long end of the swap curve are primarily pension funds and insurance companies. In this section we outline the main players, their sizes relative to GDP, their current asset allocation and their main reasons for investing in the EUR long end. Exhibit 3 shows a schematic of the different players in different segments of the swap curve. We focus on the long-end players, in particular on real-money demand from pension and insurance funds and speculative demand from hedge funds. Historically pension funds would be most active in the 3y-5y part of the curve. But recent regulatory changes being proposed under Solvency II framework are changing the dynamic in the long end. We review these changes later in this report. Exhibit 3: Sources of demand for different segments of the yield curve Short end Belly Long end Volatility Relative Value Arbitrage Opps Hedge Funds Retail Clients Structured Products Companies Hedging Investment Banks/ Dealers Retail Banks Yield enhancement B/sheet hedging Reduced funding costs Hedge Funds Pension Funds Hedging l/term liability payments Yield enhancement Public Sector Hedging Volatility Relative Value Arbitrage Opps Reduced funding costs Insurance Companies Hedging guarantees Hedging other rate risk in portfolio Exhibit 4 shows the distribution of pension fund plans across various countries and highlights the importance of this sector to the main European countries. In the Netherlands, pension funds are approximately 13% of GDP. Given the importance of the Dutch pension sector to the long end, later in this section we will review Dutch pension funds and their asset allocation. The other important point is the concentration of pension fund investment in fixed income instruments. For example in the Netherlands, close to 45% of total pension plan assets are in bills and bonds and 35% in equities. In Denmark, this is more pronounced with 7% invested in fixed income instruments. Thus changes in the way pension funds manage their investments has a profound impact on European fixed income markets. Drivers of the EUR long end 6

7 Exhibit 4: Pension funds as a percentage of selected OECD countries GDP Exhibit 5: 21 Pension fund asset allocation for selected OECD countries (% of total investment) Where data available Netherlands Switzerland Australia United Finland United States Canada Denmark Ireland New Zealand Portugal Sweden Spain Norway Austria Germany Italy Belgium France Greece Australia United States United Kingdom Finland Canada Switzerland Germany Netherlands Italy Belgium Greece Portugal Austria Sweden Spain Norway Denmark Cash and Deposits Bills and Bonds Equities Other Source: OECD Global Pension and Insurance Statistics, Credit Suisse Source: OECD Global Pension and Insurance Statistics, Credit Suisse Dutch pension funds Total Dutch pension fund assets are 86bn as of Q1 212, according to the Dutch Central Bank (DNB). Dutch pension funds are currently under-funded with a coverage ratio of 99%. The Financial Assessment Framework (FTK) states that a pension fund needs to have a minimum coverage ratio of 15%, i.e., the pension liabilities need to covered by 1.5x in capital. In addition the pension fund must hold a buffer to cope with financial setbacks once these buffers are included, the required coverage ratio increases to about 125%. Both valuations are done at market prices. If a fund falls below the required coverage ratio, it will be given a grace period of 3-5 years to improve the ratio. In general, pension schemes also have conditional indexation in principle this means that pension rights are revised annually for inflation or increases in wages. It s conditional in the sense that the adjustments are dependent on the financial situation of the fund if the fund is in a poor state, then the adjustment will not be applied. Exhibit 6 shows the different types of defined benefit schemes available and shows the clear shift away from final pay schemes. The majority of Dutch pension schemes are Defined Benefit schemes at the end of 28, 1% of active members had a final salary scheme, 87% had an average salary scheme, and 5% a Defined Contribution scheme (source: DNB). The point that is important for us in fixed income is that defined benefit schemes still remain the higher proportion of pension schemes; hence, they keep long-dated liability hedging needs high. Drivers of the EUR long end 7

8 27Q1 27Q2 27Q3 27Q4 28Q1 28Q2 28Q3 28Q4 29Q1 29Q2 29Q3 29Q4 21Q1 21Q2 21Q3 21Q4 211Q1 211Q2 211Q3 211Q4 212Q1 27Q1 27Q3 28Q1 28Q3 29Q1 29Q3 21Q1 21Q3 211Q1 211Q3 212Q1 2 July 212 Exhibit 6: Distribution of pension schemes indicate a move away from final pay schemes During the financial crisis and now during the European sovereign crisis, Dutch pension funds have reduced their equity holdings and increased fixed income proportion at the same time as decreasing their holdings of euro-denominated assets. Exhibit 7: Since the 28 crisis pension funds have migrated more to fixed income Equities Fixed yield securities Exhibit 8: % Dutch pension plan assets invested in euro assets % of assets in Euro Source: DNB, Credit Suisse Source: DNB, Credit Suisse Coverage ratios have declined sharply during the sovereign crisis, increasing the need to buy fixed income instruments. A common theme running through this report is the selfdefeating feedback loop between declining coverage ratios increasing the need to receive rates, which pushes rates lower which then acts to increase the present value of liabilities. This in turns makes coverage ratios deteriorate further. We discuss how regulatory reform and more flexible DNB rules are helping to address this. Using total assets of Dutch pension funds and their current coverage ratio, we calculate that these funds are under-funded by 5-2bn. In section Regulation will be key driver we discuss how recent regulatory changes will impact this coverage ratio and the need to hedge liabilities. Drivers of the EUR long end 8

9 Austria Belgium Denmark France Germany Greece Ireland Italy Netherla Portugal Spain Austria Belgium Canada Finland France Germany Ireland Italy Netherl Portugal Spain Sweden Switzerl Total assets (EUR, bns) % of total investment portfolio Mar-2 Sep-2 Mar-3 Sep-3 Mar-4 Sep-4 Mar-5 Sep-5 Mar-6 Sep-6 Mar-7 Sep-7 Mar-8 Sep-8 Mar-9 Sep-9 Mar-1 Sep-1 Mar-11 Sep-11 Mar-12 Total assets/insurance provisions 2 July 212 Exhibit 9: Dutch pension fund coverage ratios have deteriorated during the financial and sovereign crises Source: DNB, Credit Suisse Insurance companies Pension and insurance markets are significant players in the euro swap market, especially in the EUR long end. We outline the size of the insurance market using data from both the OECD and the ECB. At the end of this section we aggregate the data to look at these two long-end players in aggregate. Exhibit 1 shows the size of insurance company assets in the main euro area countries, and the distribution of their investment portfolios is shown in Exhibit 11. French and German insurance corporations dominate the market in absolute size we discuss the changes being proposed under Solvency II and the reduced need to hedge long-dated liabilities due to changes in the discount curve in section Regulation will be key. Yet again, holdings of fixed income are an order of magnitude higher than other asset classes across all European insurers. As shown in Exhibit 11, over 7% of total investments by French insurance companies are in bonds. In Belgium and Italy, this is equally sizeable at 8% and 75%, respectively. Exhibit 1: Total insurance fund assets in Europe (EUR, bns) % of GDP shown above the chart; Denmark, Germany, Ireland, Netherlands data refer to life insurance only 1,4 1,2 1, Insurer assets Exhibit 11: Distribution of investment portfolio skewed towards fixed income Real Estate Mortgage Loans Shares Bonds issued by public and private sector Loans Other Than Mortgage Loans Other Investments Source: : OECD Global Pension and Insurance Statistics, Credit Suisse Source: : OECD Global Pension and Insurance Statistics, Credit Suisse Drivers of the EUR long end 9

10 Austria Belgium Germany Spain Finland France Netherlan ds Total assets (EUR, bns) Austria Belgium Denmark Finland France Germany Greece Ireland Italy Netherlands Norway Portugal Spain Sweden 2 July 212 Total insurance fund assets in France total 1.2 trillion or over 65% of French GDP. In Germany, insurance company assets are also sizeable at 1 trillion. In both markets insurance premiums are also significant as shown in Exhibit 12, indicating significant flow of funds into this sector of the European economy. Exhibit 12: Gross premiums for insurance companies in Europe (EUR bns) Above % of GDP Source: OECD Global Pension and Insurance Statistics, Credit Suisse Total gross premimus (EUR, bns) Combining pensions and insurance markets in Europe In this section we include the data for both pension and insurance funds given the different ways life pensions are accounted for in some countries. For example, life pensions are provided by insurance companies in France. Thus, it makes sense to look at these two sectors in aggregate. Total pension and insurance assets in Germany are 1.98 trillion as of the end of 211 (8% of German GDP). In France, the numbers are just as significant, with over 1.9 trillion in assets under management in insurance companies. At the euro area level total pension and insurance fund assets are over 6.5 trillion or 9% of euro area GDP. Now that we have outlined the main players and their size relative to GDP, we move onto the drivers of the EUR long end. The main four factors we focus on are: 1) Traditional market factors such as level of yields, steepness of money market curves, inflation expectations, etc. 2) Dynamics of the long end during the financial crisis of ) Drivers during the sovereign crisis. Exhibit 13: Total pension and insurance fund assets are over 2 trn in Germany and France % of respective GDP shown above the bar chart 4) Going forward, the influence of regulatory reform on demand for the long end Source: ECB, OECD, Credit Suisse Pension fund total assets Insurance corporations Drivers of the EUR long end 1

11 Drivers of the EUR long end In a normal economic cycle the long-end slope is largely a function of the following traditional factors. Market factors Level of front-end yields: in normal economic cycles, curve shape is largely driven by the level of front-end yields as shown in Exhibit 14. As rates rise in the front end, we would normally expect 1s3s to flatten. During the financial and sovereign crisis, the anchoring of front-end yields at extreme lows has somewhat changed this relationship. And idiosyncratic factors, which we discuss in greater detail in this report, have also led to decreased sensitivity to the level of front-end yields. Exhibit 14: As 2y rates fall, EUR 1s3s would normally steepen -1 Exhibit 15: EUR 1s3s versus 5y5y inflation swaps Jul-2 3-Dec-4 1-Jul-7 3-Dec-9 EUR 1s3s EUR 2y rhs inv 3-Dec-9 3-Dec-1 31-Dec-11 EUR 1s3s EUR inflation 5y5y rhs Locus Inflation expectations and level of exchange rate: expectations for inflation are also a key driver of curve slope in an inflationary environment without a credible central bank response we would expect bear steepening in 5s1s which then filters through into a steeper long end. In the same vein, deflation scares would cause bull flattening in our view. Exhibit 15 shows the relationship between EUR 1s3s and 5y5y inflation swaps. Term premium is a function of forward inflation expectations which are a function of current inflation and exchange-rate risks. We proxy this with the level of 5y5y inflation swaps in this chart. Global 1s3s: Europe is also closely linked with global fixed income and any steepening in USD curve is also likely to drag the EUR curve steeper as well (see Exhibit 16). This is less of a factor in the very long end due to specific idiosyncratic flows that are specific to Europe. Long-end volatility (inter-dependent relationship): EUR 1s3s is very closely linked with gamma on long tails. There are two main reasons for the relationship between volatility and 3y rates. o Flow specific: This is partly due to legacy exotic dealer positions and hedging. It is also due to pension fund and insurance markets using the vol market and the swap market interchangeably to manage long-dated liabilities. Since the start of the financial crisis, an increase in gamma on 3y tails corresponded to flattening of EUR 1s3s. We expect this relationship to remain in place for the foreseeable future. Drivers of the EUR long end 11

12 o Mathematical relationship: There is also a mathematical relationship between volatility and the convexity adjustment. We discuss this relationship in more detail in the section on the Convexity Adjustment. Dealer hedging flows: The long end is very flow dependent. Mortgage flows and exotic hedging flows are important drivers of the long end with dealers acting as intermediaries. We discuss these in more detail in the next section. Mortgage-related flow: Historically, housing associations would be natural payers of long-end rates to hedge their mortgage books. Recent news reports highlighted the concerns that one Dutch housing association may not be able to meet margin calls from earlier this year. Unwinding these positions would lead to lower 3y yields and flattening of the curve. Arbitrage opportunities: this is primarily driven by hedge funds trying to anticipate flows from the main players (pension funds and insurance sector). Hedge funds are also quick to exploit any dislocations or positive carry and roll-down in the EUR long end caused by recent volatility. Exhibit 16: EUR 1s3s correlated with USD 1s3s 1 Exhibit 17: EUR 1s3s versus 1y3y vol Dec-99 3-Dec-4 3-Dec-9 USD 1s3s EUR 1s3s 3-Dec-9 3-Dec-1 31-Dec-11 EUR 1s3s EUR 1y3y ann bp vol rhs inv -125 Locus Regulatory factors IASB accounting changes for pensions: regulation for company pensions to remove smoothing when evaluating pension plan assets is expected to come online on 1 January 213. Within this proposal companies will not be able to assume unrealistic returns on their plan assets when calculating pension expense or the funded status of their pension plans. Thus pension funds that have large liability mismatches will face increasing pressure to accelerate liability management in coming months. Since the EU has adopted IFRS rules, this is likely to affect company pension plans in Europe. With this proposal, companies may be forced to use their discount rate as a proxy for their expected return on plan assets. Market expectations are that companies with poor funded status will have to increase their duration buying to match liabilities. This is one force supporting demand for long-dated fixed income investments. Demand exists from pension funds and insurance companies to hedge liabilities (which is linked to level of yields, equity performance, regulation Solvency II), which we discuss in section Regulation will be key. Credit Counterparty Credit Risk Management (CCRM) and Counterparty Valuation Adjustments (CVA) hedging demand for legacy positions also affect the EUR long end. Drivers of the EUR long end 12

13 Spread (bps) 2 July 212 The convexity adjustment is also a driver The convexity adjustment arises from the nonlinear relationship between price and yield. As yields fall, prices rise by more than that Exhibit 18: Convexity adjustment for swaps implied by a linear function. Also when yields rise, prices fall less than that forecasted by just a linear relationship between yield and price. This benefit of being long a bond is called the convexity adjustment. We discuss the convexity adjustment here because this adjustment has a larger impact on long-dated fixed income instruments than on the front end. The equation to calculate the convexity adjustment is shown in Exhibit 18. For a forward rate E(y) we approximate it using a futures rate, estimate the forward rate of the current curve, and then apply this adjustment. The variables are the yield on the future or par bond y, the maturity of the forward, T, and the price of the par bond or future, P. The convexity adjustment includes the ratio of the second derivative of the bond price to the first derivative. The volatility term is represented by σ we approximate the volatility using normal swaption vol of the appropriate expiry and tenor. As shown the convexity adjustment is proportional to vol squared and is quadratic in the maturity (duration) of the swap. As the convexity adjustment is positive for the long or the receiver in the swap, a swap investor should be willing to accept a lower rate than that implied by the futures curve. This explains intuitively why forward yields tend to be lower than futures-implied yields. Due to the significant convexity adjustment for long-dated swaps, forward curves normally slope downward in the long end (see Exhibit 19). In order to analyse the true implied forward yield we need to strip out the convexity adjustment. Exhibit 2 shows the EUR 1s3s curve at various forward horizons for both the adjusted and unadjusted convexity effect. As shown once the convexity adjustment is stripped out, the curve is not expected to invert at very long forward horizons. In fact the curve is expected to steepen substantially between the 5y and 2y horizons. Knowing the true implied forward curve is important when deciding whether a forward curve trade is positive or negative roll. Exhibit 19: Forward curves are typically downward sloping due to the convexity adjustment.75 Forward Swap Curve (Fixed Tenor) 25 Exhibit 2: Convexity adjusted forward rates and the EUR 1s3s (convexity adjusted) EUR 1s3s - removing convexity adjustment effect EUR 1s3s - normal swap curve Spot 2Y 4Y 6Y7Y8Y 1Y 12Y 15Y 2Y 25Y 3Y Forward Horizon EUR 3YR - 1YR EUR 3YR - 1YR 6 Months Ago Locus, the BLOOMBERG PROFESSIONAL service -2-4 spot 1Y 2Y 3Y 5Y 1Y 2Y Forward horizon Drivers of the EUR long end 13

14 The convexity adjustment can also provide an intuitive explanation of the relationship between volatility and curvature. Exhibit 21 shows the EUR 5s1s3s swap fly versus EUR 1y3y volatility. As volatility increases, curvature increases. We can now see that curvature increases as the convexity adjustment rises and disproportionately affects the long end of the curve. Thus paying a fly is effectively being long gamma and more importantly long convexity which is an attractive property. Our RV rates strategist, Panos Giannopoulos, has pointed out the latest dislocation in the very long end shown in Exhibit 22. With discussion of the pension fund regulatory changes, the EUR 2s3s part of the curve has steepened sharply, pulling the EUR 5y, 1s15s3s fly to all-time rich levels. This chart shows that the fly has moved out of line with volatility volatility (and hence the convexity adjustment) have increased while the fly has richened substantially. The 5y, 1s15s3s fly is at extreme levels. In our view, although this makes sense over a multi-year horizon, the fly has moved too quickly lower, so we expect some flow to fade this move in the short term. Exhibit 21: EUR 5s1s3s curvature linked to the level of volatility Exhibit 22: EUR 5y, 1s15s2s fly versus volatility Jul-2 3-Dec-4 1-Jul-7 3-Dec-9 EUR 5s1s3s, Spot forward EUR 1y3y daily bp vol Locus In the analysis below we divide the discussion into two periods: during the financial crisis which we define as the period between 27 and March 29 and the current period of the sovereign crisis (29-now). We discuss the flows during the financial crisis and how they are different to the current period, for example, exotic dealer desks were much more active in the swap curve than is presently the case. Drivers of the EUR long end 14

15 Drivers during the financial crisis The EUR 1s3s curve inverted at the height of the financial crisis in 28 (Exhibit 23). This was mostly driven by exotic dealer hedging flows, expectations of a prolonged recession, and pension fund receiving flows. As the market priced in the probability of an upcoming recession or downgraded growth expectations, the entire term structure flattened. In the process, the 1s3s curve flattened sharply and at one point the 1s3s curve inverted. Exotic dealer positions in the long end were also significant contributors to the reshaping of the yield curve in We discuss this in detail below. Exhibit 23: EUR 1s3s inverted during the financial crisis Locus Exotic gamma hedging and the gamma trap Exotic vol desks had sold popular CMS spread-linked notes that benefited investors as long as a particular curve segment was non-inverted. These included CMS spread range accrual notes (CRANS) where the payoff was typically paying a high coupon, multiplied by the number of days in a coupon period the curve was not inverted, divided by the total number of days. This was one popular play for non-inversion of the EUR 1s3s curve. CMS spread range accrual structures leave dealers short linear floors and long digital floors on the CMS spread between the two rates referenced in the notes. The digital floor, or put, is struck at zero CMS spread that is, the dealer does not pay coupon on range accruals if the curve inverts. The short linear floor exposure arises from the fact that dealers are paying investors in CMS spread leveraged notes a linear factor multiplied by the CMS spread, floored at zero. The issue arose when the curve turned negative, which left dealers significantly short gamma. In this situation, instead of doing the opposite of the market, which would have resulted in hedging needs opposing the trend in swaps curve dynamics, dealers needed to hedge in the same direction as the market, reinforcing the trend. This pushed 1s3s flatter in a rallying market. Thus many exotic dealers had to rehedge via receiving the long end, which further exacerbated the inversion of the curve. This meant that dealers went from exposure being relatively flat gamma in a steep yield curve environment to significantly short gamma in an inverted curve environment this caused the market to gap with increasing volatility and was referred to as the gamma trap. Drivers of the EUR long end 15

16 Mar-2 Sep-2 Mar-3 Sep-3 Mar-4 Sep-4 Mar-5 Sep-5 Mar-6 Sep-6 Mar-7 Sep-7 Mar-8 Sep-8 Mar-9 Sep-9 Mar-1 Sep-1 Mar-11 Sep-11 Mar-12 2 July 212 As the crisis progressed and evolved from a banking crisis to a sovereign debt crisis, dealers were able to roll off or sell the long gamma positions on the 1s3s curve. This served to reduce the effect of exotic dealer hedging on the long end, and volatility in the long end subsided. We still think some of these legacy CRAN positions exist but in much smaller size than during the financial crisis. Thus, going forward we expect exotic dealer hedging of these CRANS to be less important in driving the EUR long end. In contrast, pension fund and insurance company hedging needs, discussed below, are likely to remain dominant drivers for the long end. Pension fund asset-liability management During the financial crisis pension funds and insurance funds were held hostage to falling long-end yields due to the crisis and the gamma trap. As equities fell and yields moved lower, pension fund solvency ratios fell (Exhibit 24). To hedge their asset/liability mismatch, pension funds needed to receive more in the long end. Pension funds were particularly affected during the financial crisis by inversion of the curve and saw their cover ratios plummet. This flow has followed through from the financial crisis into the sovereign crisis and remains a key driver of the EUR long end currently. The main innovation from regulators during this period was to provide some breathing space for pension funds by giving them time to return to the required cover ratios and easing the burden of liability management in a falling yield environment. This topic is now in full focus in the market as we have reached new lows in long-end yields. We discuss recent proposals for pension fund reform in greater detail in the section Regulation will be a key driver. Exhibit 24: Dutch pension fund cover ratios versus Equities Equities - Eurostoxx 5 2 Dutch Total assets/provisions (Cover Ratio, rhs) Source: DNB, the BLOOMBERG PROFESSIONAL service, Credit Suisse Drivers of the EUR long end 16

17 Drivers during the sovereign crisis When the sovereign crisis in Europe flared up, long-end dynamics were then largely driven by pension fund receiving demand and CVA-desk (counterparty valuation adjustment) hedging their legacy peripheral sovereign exposure via swaps. CVA desks monitor counterparty credit risk for a bank s OTC positions and manage the exposure of a loss if a certain counterparty defaults. The risks of default and the fact that sovereigns are no longer risk free are so important now following the Lehman collapse and the credit event in Greece. At a very high-level exposure in OTC positions needs to be adjusted by default probabilities of the swap counterparty. Intuitively, a valuation adjustment is done by shifting the discount curve by the CDS spreads and then calculating the PV of any positive-valued swap contract. Thus, if a bank has a profit with respect to a counterparty, the PV of that position needs to be reduced to account for the chance of counterparty default. CVA hedging flows As CSAs (credit support annexes) with sovereigns are not always collateralized, banks face credit counterparty risk when banks positions move into profit versus the sovereign. CSAs are normally one-way with a sovereign so that when a sovereign is in profit on a position then the swap counterparty has to post collateral, but the reverse is not necessary. Banks are normally in profit versus the sovereign in a rallying market, which ironically coincides with widening peripheral spreads. When peripheral spreads widen, dealer desks then need to manage counterparty credit risk and this normally means receiving the long end of the swap curve and buying protection in CDS markets. Exhibits 25 and 26 show that the 1s3s curve has been linked to movements in peripheral yield spreads and CDS. These charts show that the EUR 1s3s curve would flatten during periods of peripheral spread widening. In our view, this is partly related to CVA hedging. Also, the apparent outperformance of swaps versus bonds in periods of stress also indicate that CVA hedging flows are predominantly in long-end swaps (Exhibit 26). Exhibit 25: EUR 1s3s versus Italy/German 1y spread 4 Exhibit 26: Italian sovereign CDS versus German 3y ASW spreads Dec-1 1-Jul Dec-11 EUR 1s3s italy vs germany 1y rhs Locus 1-Jul-1 3-Dec-1 1-Jul Dec-11 German 3y ASW Italy 5y CDS rhs Drivers of the EUR long end 17

18 CVA desks hedge both in the credit market via CDS and via the rate market. Given that the European authorities allowed a CDS trigger event when Greece restructured its debt, we think CVA desks will be more comfortable hedging their positions with CDS. One point to bear in mind is that ongoing CVA hedging demand can have both a steepening or flattening effect on the EUR long end depending on the performance of peripheral spreads. Recently with peripheral spreads tightening, paying interest from CVA desks has contributed to steepening pressure in the long end. Thus reduced vol in the periphery should support further steepening of the curve. We also expect banks to slowly start introducing two-way CSAs with sovereigns which should also lessen the impact of sovereign CVA hedging on the EUR long end. We have already seen increased funding costs pushing the Bank of Portugal and the Bank of Ireland to introduce two-way CSAs and the recent announcement by the BoE also highlights this gradual shift towards two-way CSAs with sovereigns. Pension fund asset-and-liability management flows remained a key driver of the long end during this phase of the crisis. As equity market performance deteriorated during the crisis, pension fund solvency ratios fell further. In order to hedge liabilities, pension funds needed to receive long-dated rates, exacerbating the long-end flattening. In the next section we move onto regulatory reform and how we expect this to drive demand for the long end going forward. Drivers of the EUR long end 18

19 Regulatory changes in the pension and insurance sector are key Solvency II demand for the long end shifting to 2y point Solvency II is new regulation to ensure that European insurance companies hedge liabilities to reduce risk, just as pension funds do. Exhibit 27 shows the pillars of the Solvency II proposal. Regulators are still in discussion about how to value liabilities under the Solvency II proposal, and this has implications for demand for the long end of the swap curve. Exhibit 27: The three pillars of Solvency II all three pillars are being developed concurrently Pillar 1 Pillar 2 Pillar 2 Pillar 3 - Balance sheet evaluation - Solvency Capital Reqt (SCR) - Minimum Capital Requirement (MCR) - Investment rules and internal models - System of governance - Own risk and solvency assessment - Supervisory review process - Publication of annual solvency and financial condtion report to the public and regulator - Report to include valuation basis, risk and capital mgmt process Quantitative requirements Qualitative requirements Disclosure requirements Omnibus II, part of the Solvency II regulatory proposal, was voted on by the EU Parliament on 21 March. Although the proposals in the Omnibus II Directive passed the current phase, they are still subject to further approval at the trialogue level before being finalized (between the European Commission, The European Council, and the Parliament). The next date will be the plenary vote on Omnibus II by the European Parliament unfortunately the date remains in flux due to lack of agreement on the current proposal. It is possible (and likely) that this is pushed to a post-summer vote in September. Unfortunately, the discussions remain private until full approval is reached; thus, we can only indicate market concerns and our expectation of changes: Discount curve and ultimate forward rate The first phase of Solvency II proposed the use of the 3y point as the last market rate to use and thereafter we had expected extrapolation to the 6y point towards an ultimate forward rate of 4.2%. It now appears as if the extrapolation will happen between the last liquid point set at 2y and 1y hence. Effectively this means that the extrapolation to an ultimate forward of 4.2% happens for all liabilities between the 2y and 3y points. This would act to reduce volatility in the present value of long-dated liabilities, but more importantly would reduce the value of these liabilities related to original expectations. But it is still important to bear in mind that debate about the methodology on the Ultimate Forward Rate is still ongoing. The European Commission and the EU Council favour convergence to the UFR in 4y after the last liquid point of 2 years. But the European Parliament prefers faster convergence within 1 years (similar to the Danish proposal, see later). Drivers of the EUR long end 19

20 If no agreement is reached soon then it is more likely that the market will have to wait for September for further information. But the broad point for our markets is clear to us, the adoption of the 2y as the last liquid point on the swap curve will likely support further richening of 2y versus 3y and the ultra-long end. By proposing that liabilities longer than 2y are discounted at a rate extrapolated from the rate curve to an ultimate forward rate (UFR) of ~4.2%, the EU regulators are effectively reducing the liabilities that insurance companies need to hedge for regulatory purposes. With interest rates at new lows, pension funds in Europe are also trying to converge to the Solvency II methodology quickly we discuss the existing proposals in Denmark, Sweden, the Netherlands in more detail in the next section. Exhibits 28 and 29 illustrate how the discount curve can change under the new Solvency II proposal. We show the current swap curve, the projected curve under the scenario where the 2y is the last liquid point and convergence is expected 1y after (shown as 2-3 curve) and the projected curve out to 2y with convergence achieved 4y hence (shown as the 2-6 curve). As shown there is a substantial difference in the discounting curve past the 2y point. Exhibit 28: Illustrative spot discounting curve 3.5% 3.% 2.5% 2.% Exhibit 29: Illustrative 1y forward discounting curve 5.% 4.% 3.% 1.5% 1.%.5%.% Spot rate 2-6 Spot rate Current spot curve Maturity 2.% 1.%.% Spot rate 2-6 Spot rate current 1y forward curve Maturity Change in discount curve reduces liabilities by 1%-15% We now illustrate how the present value of different liability structures will change depending on which discount curve is used. We consider three liability structures: 3y bullet, 5y bullet and a sample 6y liability profile (seen in Exhibit 3). The table in Exhibit 31 shows that if a 5y bullet liability is discounted using the 2-6 curve, the PV of the liability is only 7% of that using the existing swap market rates, i.e., there is a 3% reduction in the value of the liability. This is further reduced to 62% when using the faster convergence curve, the 2-3 curve. For the more realistic 6y liability profile (shown in Exhibit 3), we find that the change in discount curve to the 2-3 curve reduces the PV by 13%. This gives us an idea of the reduced need for hedging liabilities going forward for both pensions and insurance markets. If we assume that this sample liability example is representative of the Dutch pension fund liability structure, then we estimate that this would reduce liabilities by 13% and decrease hedging needs by about 5-1bn. Drivers of the EUR long end 2

21 EUR (mns) 2 July 212 Exhibit 3: Illustration: long-term liability cash flow illustration liability cash flows Exhibit 31: Change in PV of a liability due to change in the discount curve % of PV under a normal swap curve PV 3y bullet liability PV 5y bullet liability Sample 6y liability normal swap curve y y , the BLOOMBERG PROFESSIONAL service The timeline for Solvency II June 213: Acceptance of Solvency II by trialogue. Legal requirements will need to be transposed into national law by 1 June 213. This was originally set for the 1 January 213, but due to delays with the Omnibus proposal, the date has been shifted forward. January 214: implementation date of Solvency II rules by industry participants. This is the start date for the phasing in of implementation for industry participants. Recent press reports suggest that the EU is considering phasing in the new capital rules over a seven-year timeframe. For the phase-in period of seven years, insurance companies would be allowed to use the existing discount rate under Solvency I while transitioning to the market discount rate. A schematic timeline is shown in Exhibit 32. Exhibit 32: Schematic timeline for adoption of Solvency II March 212 Omnibus II is adopted in ECON on 21/3/212 April June 212 Trilogue on Omnibus II between EP (ECON), Council, and Commission September 212 Possible adoption Omnibus in the EP November 212 Proposal Level II December 212 Original plan: EUC to issue a Delegated Act (regulation) with L2 measures January 214 Insurance companies implement Solvency II January 212 Danish presidency until 3/6/212 April 212 Scheduled adoption of Omnibus II in plenary July 212 Cypriote presidency until 31/12/212 October - November 212 Omnibus II published in OJ 31/1/212 SII directive 29/138: date of entry into force of SII directive 3/11/212 SII directive 29/138: date of application of all SII directive requirements (and repeal of Solvency I December 212 EIOPA to consult on draft technical standards 31/12/212 Proposal O II: Entry into force of SII directive (member states deadline for transposition) June 213 Transposition of Solvency II into local regulation completed Drivers of the EUR long end 21

22 Pension fund reform converging to Solvency II In this section we discuss recent announcements by European and Nordic pension authorities that are also impacting the EUR long end. Announcements by the Dutch, Danish, Swedish, and Finnish pension authorities over the last few weeks led to a sharp sell-off across the EUR curve, with 1s3s bear steepening by over 1bp. As we discussed in EST, 1-Dec-11 and EST 2-Feb-12, 1s3s steepening has been one of our main core views for 212. We argued that the proposals under Solvency II support the steepening in the long end. Now, the acceleration of pension fund changes further corroborates this steepening view. Low rates forcing regulators to take emergency measures Pension funds and insurance companies have been held hostage by the fall in interest rates. Their hedging needs further exacerbate the negative self-fulfilling spiral between falling equities and falling interest rates. Pension regulators acknowledged this selfdefeating policy, which ultimately makes pension plan members suffer in the short term and hence forces them to step in. To be fair, most proposals were flagged to the market before the surprise was the quick turnaround time between proposal and agreement. Below we summarise the main measures announced in the last few weeks: Sweden The Finansinpektionen introduced a voluntary floor on the discount curve due to exceptional market conditions. The regulator effectively gives life insurance companies and occupational pension funds the ability to use the discount curve derived from the Swedish interest rate curve as of the 31 May 212 as a floor for discounting liabilities. The proposed floor is effective from the 3 June 212 and is expected to apply for one year. In our view, the risk of this temporary measure is that it may evolve into the status quo if rates continue to rally. As evidenced by a similar announcement in September last year, these measures have a way of being extended and extended. In the short term this is undoubtedly positive for pension fund cover ratios, but we question the long-term viability of this strategy. Denmark The Ministry for Business and Growth signed an agreement with the association of pension and insurance companies to lift the discount curve equivalent to a more ordinary market condition. The main reason cited was to allow these industries to take a longerterm perspective in investment decisions and reduce the need to lock in future returns at the current level of low yields. The other important comment made by the regulator was that this change in the discount curve methodology would be consistent with the proposals under Solvency II. One should not forget that under Solvency II, the Danish discount curve was supposed to use the UFR starting from the 2y point out to the 3y point so this is not new information. The surprising point was the Danish pension regulators agreed to adopt this for pension funds way ahead of implementation of Solvency II (SII is still being discussed and is only expected to come online in 214). We see this as more a function of market volatility and the extremely low level of yields. Thus any move lower in yields from here may force other pension regulators down the same route once again supporting the steepening of the EUR 1s3s curve. Drivers of the EUR long end 22