Today, Chris Whalen's Institutional Risk Analytics carries a fantastic piece by Alan Boyce, in which the author picks up where we left off some time ago in deconstructing the DV01 of the Federal Reserve's SOMA. As a reminder, using Jefferies data, we observed that the Fed's DV01 on its balance sheet is about $1 billion (the potential unrealized profit/loss for every basis point move in interest rates, and with ZIRP here, rates can only go up, so make that just loss without the profit) . Alan Boyce, a former Fed member, CFC executive, and Soros portfolio manager, provides a more granular analysis of the Fed's holdings and comes up with an even scarier DV01: one that is 50% higher, or a $1,509 million/bp. This means that the Fed faces a "$75 billion loss for the first 50 bps move in the markets." As before, it is obvious why the Fed will do everything in its power to keep rates as low as possible for as long as possible, as the vicious cycle that will begin with increasing rates will make all future press releases of how much money the US taxpayer has "made" on the US' bailout of the mortgage industry far more problematic. Boyce also discusses how precisely it is that the Fed has managed to maintain rates at current record low levels for so long, what the cost of an appropriate hedging portfolio would be, and, critically, the implications of what will happen when markets realize that we are caught in a state of artificial suspended and Fed-endorsed animation. The primary conclusion: look for interest rate volatility to surge by 50% even as the Fed scrambles to cover hundreds of billions in losses in its portfolio sooner or later. Boyce's summary is basically that the countdown to the end of the Fed QE regime is now on: "If you look at forward fed funds (Eurodollar curve less basis swap), the FRB will go negative carry in March 2015, where 3 month financing rates are forecast to be over 5% (just gets worse and worse from there). The point here is that mark-to-market accounting is an iron law. You cannot escape the losses just because you do not report them. If the FRB loses $200 billion on mark to market, there will be $200 billion LESS that they remit to the Treasury Department every year. That will require legislation to either raise taxes or lower spending by $200 billion (or run up bigger Federal debt to be paid back by another generation)." Must read analysis.
First, a summary of the Fed's actions from the perspective of an MBS trader:
Over the last year the administration and the Fed have undertaken several measures, in the form of FPP, that are yielding short term positive benefits. Specifically:
1. Duration - the FRB/Treasury's programs have taken more duration out of the bond market than could be created in 2-3 years. This has kept long term interest rates much lower than they would otherwise have been. The yield curve has been artificially depressed, by at least 75 basis points, due to this reduction in the aggregate duration (price risk of the whole US bond market). This directly impacted credit and increased the price of all credit instruments.
2. Reduced Volatility - The result of the FPP removing a sizable chunk of mortgages out of private market hands is a reduction in the amount of convexity hedging in the near term. This drives down actual volatility in the fixed income markets. The FPP also forced many bond fund managers to replicate mortgages, in part by selling volatility, thereby compressing implied volatility. This became self-reinforcing as lower volatility allows for increased risk in portfolios which in turn reduces volatility as liquidity returns to markets. Volatility compressed across asset classes. And, reduced volatility directly impacted credit spreads, equity valuations, and commodity prices.
As we know, the FRB recently ended its purchase program. And this is where the mortgage trader's perspective may prove useful. A clear and consistent focus on mortgage duration risk shows us that we may face a series of large and as yet underappreciated challenges. The sooner we collectively understand these issues, the better we can address them. The rest of this article will describe my efforts to calculate the duration of the FRB/Treasury purchase programs and give my perspective on what will happen when the purchase programs end and underlying dynamics of mortgage duration assert themselves.
How does one calculate the specific DV01 of the Fed and how much will a hedge cost?
Annex 1 shows the MBS position of the FRB as of 12/30/09, which includes 'reported' settled positions and forward purchases (reported weekly). Essentially, this is the risk position of the Federal Reserve. An MBS trader (or investor) would calculate their risk using such a position sheet, in combination with various estimates of the duration of each underlying security. All of the analysis is done using 12/30/09 prices and durations.
By using OAD calculations from standard prepay models (which are too fast), the FRB had $460 million of price risk per basis point or "mm/bp" as of year-end. If you scale this up for the remaining purchases and the purchases by the Treasury Department, you get $776 mm/bp. This is using Option Adjusted Durations implied by pre-payment models that were conditioned on the housing market from 2002 to 2008. Those prepayment estimates are significantly faster than what has been experienced in the last two years.
If you use durations implied by coupon spreads (a pretty accurate measure of how the bond market views the current price risk on MBS) then your duration value of a basis point move in the markets or "DV01" jumps to $643mm. Scaled up for the whole program, you get $1,071mm/bp. If rates go up by 50bp, the FRB and Treasury would expect to lose $54 billion. If rates go up even more, assume durations extend and the losses on the next 50bp increase would be $75-90 billion.
If the FRB and Treasury were to hedge their negative convexity risk (bonds fall more than they rise for a given move in interest rates) they would need to buy interest rate options. The most likely option would be a 3yr into 7yr swaption, which currently trades at 5 points. If you hedged to coupon spread implied durations, you would be buying 3yr into 10yr swaptions, which currently trade at 6.5 points. There would be significant reflexivity if that amount of swaptions were bought (prices would be higher if there were more buyers). I estimate the average price paid would be at least 50% higher and have confirmed this with some of the best option traders in the world.
Bottom line: it would cost $142 billion for the FRB and Treasury to hedge the short optionality of their current MBS position.
The Risk of Other Bonds Purchased: Treasuries, Agencies, TIPs and Maiden Lane
These calculations are again based upon year-end positions. The durations are estimated by breaking each category of debt instrument into several buckets, estimating a duration for the bucket and then calculating a simple weighted average.
Asset
Position $B)
Duration
DV01 ($M)
Treasuries
707
5yr
353
TIPs
47
6yr
28
GSE debt
160
2yr
32
Other (Maiden Lane etc)
50
5yr
25
Total
964
4.55yr
438
The weighted average duration is 4.55 years with a DVO1 of $438mm per basis point. I will assume that since December, the remaining purchases have been in MBS instead of the other debt categories. Together with the scaled up to final size purchases of MBS, that would be $1,509 mm/bp or a $75 billion loss for the first 50bp move in the markets.
So what is the real duration of the market, or in other words how much duration has the Fed taken out of the market?
Accounting for the Net Duration Add to the US bond market
Mortgage market duration is estimated to have been roughly the same during the period of the FPP. There was no large scale refinancing, a sure-fire method to increase duration. The duration embedded in the existing mortgages increased slightly due to lower housing turnover and labor mobility. The recent GSE buybacks of >120 Day delinquent loans acts to reduce the duration of mortgages, as loans that were completely unable to voluntarily prepay are removed from the system.
Municipal market shrank in 2009. This was aided by help from the Federal government in the form of Build America taxable bond issuance and significant grants to State HFAs. Corporate bond market is small and did not grow in 2009.
Net Treasury issuance as $1.4 trillion, with a 4 year maturity and a 3 year duration. This is a net add of duration of which is $420mm/basis point -- the net result. The FPP took out $1.509 billion of duration per basis point. The mortgage market, municipal market and corporate bond market are estimated to have added zero net duration to the aggregate. The funding of a very large budget deficit required a significant duration add, $420mm/bp, by the Treasury Department.
Conclusion: the FPP reduced aggregate duration in the financial markets by almost 3.6x the duration that was added to the system during the period! This has kept long term interest rates much lower than they would otherwise have been. Without this, interest rates would have been higher, the yield curve would have been significantly steeper, and options would have been more expensive.
Think the Fed's departure from the capital markets on March 31 is priced in? Think again. Market are simply once again demonstrating that EMT is totally flawed.
What are the Implications of Ending the FPP?
1) Duration: MBS widening out will result in lower prices for agency pass - thus, this will lead to an immediate increase in the calculated OAD of the aggregate mortgage index and drive the curve steeper. A steepening of another 50bp will cost the FRB another $84 billion. Agency MBS spreads are 90bp tight to their historical average spread of 120 basis points to US10yr. If spreads widen out to the average (ceteris paribus) the FRB will lose $147 billion.
Prepayment models used for convexity hedging are slow to adjust. Like historical models that failed during the crises of 2008, prepayment models will ultimately be seen as failing to recognize the enormity of the duration problem. Moving forward, refinance activity will surprise to the downside. Mortgage rates are coming off record lows. And, the creditworthiness of the delinquent homeowners still working through the system will impede refinance activity.
2) Options Volatility: The end of FPP shorting options will drive interest rate volatility up by 50%, making the cost of covering the short options position rise to $213 billion. Options traders tend to be agnostic as to what options markets they play in. When fixed income implied volatility increases, option sellers will be more likely to short options to the bond market and less likely to short options to the commodity, equity and foreign exchange markets. These options markets are all linked by investors. Expect implied volatility in all other markets to rise when the world's biggest sell program of long dated options ends.
3) Fiscal Policy: If the FRB were to just explicitly short payer swaptions, they would generate significant option premium which they would book as income. That income would revert to the Treasury department and the FRB would be wishing, hoping and praying that interest rates never move. Instead they are implicitly shorting the options through the unhedged purchase of MBS, generating cash income, which they report and remit to Treasury. If interest rates were to rise, the yield curve to steepen and/or interest rate volatility to rise, the FRB would suffer a huge mark-to-market loss. This would not be reported on their cash basis income statement. They would continue to book cash income, as long as the book yield of their MBS purchases exceeds their financing rate (paying interest on excess reserves). The Federal Reserve does not mark to market, instead runs a "cash income statement".
If you look at forward fed funds (Eurodollar curve less basis swap), the FRB will go negative carry in March 2015, where 3 month financing rates are forecast to be over 5% (just gets worse and worse from there). The point here is that mark-to-market accounting is an iron law. You cannot escape the losses just because you do not report them. If the FRB loses $200 billion on mark to market, there will be $200 billion LESS that they remit to the Treasury Department every year. That will require legislation to either raise taxes or lower spending by $200 billion (or run up bigger Federal debt to be paid back by another generation).
4) Excess Reserves: The end to the FPP will not change excess reserves, but who cares since they are being lent back to the FRB at federal funds target rate of 25bp. The Federal Reserve has plans to remove the excess reserves via reverse repo agreements (used to be called matched sales back in the non-borrowed reserve targeting regime). If the Federal Reserve wished to sell all of the bonds purchased, the expectation is that they would receive lower prices than they paid. The FRB would then need to issue "FRB bills" to soak up the remaining excess reserves, equal to the dollar loss on the round-trip bond trade.
And the one question everyone is asking: "when?"
When Will This All Unfold?
Using efficient markets hypothesis, this information is freely available so the effects of the end of the purchase program should be fully priced into the financial markets. So why didn't bonds fall dramatically in anticipation? There are a couple of potential explanations. First, financial markets are not efficient. They are not the best prediction of what will happen in the future. As recent events confirm, financial markets are quite myopic, able to see at most three months ahead.
And markets can be distorted when participants are incentivized by factors other than profit. The Fed's participation in the bond markets was not driven by a desire to profit from their purchases. They systematically purchased mortgages (and Treasuries) without regard to price. This can have a distorting effect that is not recognized until their influence on the markets is removed.
Second, their involvement in markets not only influences the prices of assets but they influence the behavior of other market participants. As discussed above, many real money buyers synthetically created mortgages because of the lack of supply due to FPP.
Finally, the Fed removed a huge stock of mortgages from the pool of available mortgages in the secondary market. The impact will be felt for some time until the flow of new mortgages begins to trickle into the secondary market providing additional liquidity. In the meantime, there is a real danger of a buyer's strike as participants sit back and wait to see what happens now that the program is finished. This is happening at a time that the economic data is coming in strong adding to pressure in fixed income markets.
If markets were to become unglued, the Fed may purchase more mortgages and Treasury debt. The question is how other central bankers and market participants would react to this. Foreign central bankers will likely snap and become sellers if the Fed decides to monetize more debt. Also, domestic and foreign market participants would likely take it as a sign that the Fed is politically unable to exit the mortgage market and unable to exit quantitative easing. As FPP ends, there is the real potential for unintended consequences in domestic and foreign markets.
To all those who say buy the market - good luck. All we are doing at this point is literally shuffling the deck chairs as the Fed has managed to plug the leak for at best 5 years. Those who have access to the discount window and to the steep yield curve have at most 5 years in which to transfer as much future assets to the present discounted at the ridiculous zero interest rate which the Fed has so far successfully managed to defend. Forget the Fed selling assets: the Fed now realizes, as we have long claimed, that the conviction move by the market to an increasing rate exposure is all that will take to destroy the balance sheet of the biggest bank in the world, that of the Federal Reserve. Various tests by the 10 Year of the 4% resistance have so far been driven primarily by risk reallocation with equities. As Boyce points out, this is untenable and is a function of the ongoing delusion between stocks and bonds that without the Fed cutting duration as much as it has (not to mention that the average UST maturity profile is still woefully short), we would now be starting at an S&P level far lower than the infamous 666 lows. Yet the inevitability of this happening sooner or later is guaranteed: unless the CNY manages to become the reserve currency in the next 5 years, and the Fed somehow succeeds to convince the world that devaluing the dollar is what the Fed is all about, all we are doing is waiting for the hull of the USS America to fill with ice cold water and slowly sink.
No comments:
Post a Comment