Monday, December 17, 2012

Trade total return swaps like Gretchen Tai: a follow up on the HP Pension Fund post



In my last post I detailed the very fine performance by Gretchen Tai, the asset manager for the Hewlett Packard defined benefit fund. I was trying to work out how it was achieved.

Being ignorant in these matters I did not even know the defined benefit fund had to file detailed accounts at the Department of Labor on form 5500.

My readers are smarter than me. So they pointed me the right direction.

Now we can work out in much greater detail how the returns were made.

The pension fund is invested in a "Master Trust". Here is a statement of the assets and liabilities of the master trust:



Its pretty clear. There are $2.9 billion in US Government Securities, $3.2 billion in corporate bonds, $2.1 billion in common stock as well as a whole lot of other things ($0.3 billion in common collective trusts, $1.6 billion in limited partnerships and venture capital funds, $0.7 billion in registered investment companies, $0.6 billion in 103-12 entities and some derivative assets).

There is also some securities lending collateral which they are obliged to return.

The next page gives net income for the Master Trust by category.



We know now the government bonds appreciated by 157 million, the corporate bonds by 104 million and interest was earned of 167 million. That appears to be the bulk of the return on the $6125 million bond portfolio. That is about 7 percent - roughly the return I might have expected from a diversified bond portfolio provided they held some duration.

Much better returns are obtained from the limited partnerships and venture capital funds. They had a return of 263 million on 1383 million in starting capital. That is a very sweet 19 percent. Gretchen Tai chose funds well.

But the eye-popping number is the return of $543 million in swaps and other derivatives. This is after a more modest 152 million gain. This woman swaps cash flows amazingly well.

Breaking up the swaps return

The accounts are kind enough to give us a list of derivative exposures (at least in some aggregated form):



There is $2.8 billion up from $2.5 billion in total return swaps.

They are also kind enough to tell us the P&L for each of these types of derivatives (none of which are designated as hedges and hence all of which are marked to market).




Those total return swaps made $561 million - 22 percent return on the starting derivative asset. Its pure alpha too - the fund needs to pay out the return on the (similar sized) liability they swapped.

And so now we know at least of part of Gretchen Tai's secret. She is the wickedest, meanest and most effective trader of total return swaps I have ever seen. A good proportion of my readers would want to hire her immediately.

So here is an aspiration in life: work out what she does - learn to trade total return swaps like Gretchen Tai. (Or just hire her.)






John

35 comments:

  1. Any disclosure of the counterparty?

    ReplyDelete
  2. John, the total return swaps were most likely just floating for fixed swaps inorder to gain duration for the liability hedge. They made so much money simply because rates fell so far down. If I had to guess she hired a 3rd party manager for the trades through a seperate account. Google LDI and youll find a ton of info on this stuff.

    ReplyDelete
  3. Banks, who thanks to the lenient Fed, have a capital cost of next to nothing, are only too happy to enter into swaps with sound counterparties.

    They borrow from the Fed at close to 0, buy securities, then enter into the swap for a fixed payment of anything above their cost of capital with the counterparty.

    There is little (in practice there is zero) risk to the bank, just pure profit on the spread between their cost of capital and the payment for the swap.

    Meanwhile, HP (or any other strong counterparty) gets levered exposure.

    In a high interest rate environment this would likely not work out nearly as well for the swap buyer.

    This is one of the very rare times such a deal is a win-win for the bank and the buyer.

    ReplyDelete
  4. They are NOT fixed duration swaps. The company reports interest rate swaps separately.

    J

    Buyers Strike - it may be a cash versus asset total return swap - but you need to find me the asset pool that returned 21 percent.

    And you need to explain why they would be so good a picking such an asset pool for the swap financing but not be any good at picking equities and debt directly.

    J

    ReplyDelete
  5. Hello everybody,

    I do not have answers to John's question but I have a question on something I do not understand : the level of receivables and payables to brokers on securities sold and bought. It looks like respectively 4% and 6% of assets (both 2010 and 2011) : either the portfolio turnover is huge or the settlement cycle is particularly long...

    Or is there a rational explanation I missed (maybe the way they are financing the trading)?

    ReplyDelete
  6. John-

    Fixed for floating = interest rate swaps
    Fixed income overlay (i.e. gaining exposure to the Long Government index for example at the LIBOR rate) = Total return swaps

    HQ's own definition "HP also uses total return swaps and, to a lesser extent, interest rate swaps, based on the equity and fixed income indices"

    The giveaway... a TR Swap on the Long Government Index was ~20% for the period October 2010 - October 2011 and ~10% from October 2009 - October 2010 (i.e. half), which basically match the gains in 2011 and 2010.

    No pension plan would give an individual the power you for some reason feel Gretchen has. This is a funny obsession you have. There are small plans out there that went 100% long bonds because the parent company didn't want to think about the pension plan that have 20% annualized performance going back to 2007 that are overseen by people in HR with some consultant oversight.

    ReplyDelete
  7. As a side comment....

    The fact that you (and other smart people) view this outperformance as being due to her is one of the reasons corporate (and public) plans are so scared of hedging their liabilities. They realize if they hedge (i.e. reduce risk of their plan) and rates rise, the fact that they outperformed their liabilities won't matter to those that see they "performed poorly" in absolute terms.

    In other words, if rates had gone the opposite way over this period, you probably would have posted "Gretchen Tai... Possibly the Worst Fund Manager You've Never Heard Of"

    ReplyDelete
  8. They SEPARATELY disclose their profits on interest rate swaps.

    So unless they are counting 2.5 billion of interest rate swaps plus another 2.5 billion of interest rate swaps that are separately disclosed as total return swaps then these TRS are NOT interest rate swaps.

    But hey - none of this disclosure is clear.

    The quote that Jake mentions also distinguishes between interest rate swaps and total return swaps.

    J

    ReplyDelete
  9. I'm not seeing inflation swaps disclosed anywhere else, and if you had shifted your DB pension fund to a liability-driven basis you would surely have needed to buy a boatload of inflation protection.

    ReplyDelete
  10. Its fairly clear you would have made a bucket load on inflation protection. The tips are all trading with negative yields.

    ---

    There is little evidence that they did that - however if they did it was

    (a) inspired and

    (b) a pretty close to full explanation.

    Would love to know. The bond portfolio - which is disclosed - is not long on tips.

    J

    ReplyDelete
  11. John:

    It could be much easier to hedge your inflation exposure via inflation swaps than buying TIPS or whatever. That said, I'd expect inflation swaps to be rolled into IR swaps, not TRS.

    Re TRS: it's hard to say what they are, but I can imagine a whole slew of what could be done. Say longevity TRS (which I know some pension funds bought, but doubt HP since it's very specialised market), and I believe those would appreciate significantly with low LIBOR as they are the ultimate in duration.

    In fact, if they were extremely fortunate and did their liability swaps via TRS in 2007/2008, then I'm suprised their returns are only this small.

    All of the above is caveat emptor as TRSes are really a huge catch-alls for any sort of shenighans.

    That said, in most cases all of the derivs will be under some sort of CSA (I can't think of any pension trust which would not require that), which indicates that the mark is not just invented by Tai but actually cash posted by (presumably a reasonably sophisticated) cpty.





    ReplyDelete
  12. Longevity swaps are one of the best - it is legit - explanations I have heard.

    Its an amazing return -

    J

    ReplyDelete
  13. yeah, looks like an asset liability hedge executed in swaps (as 99% of them are).

    So the gain in absolute terms was just a hedge of the liabilities risk. Goes to show how much liability risk most portfolios run and why the managers are so petrified.

    ReplyDelete
  14. Things like this Risk article suggest that a lot of stuff which you or I might consider to be basically interest rate derivatives business are classified as TRS by the trade, so I think Jake above might have it right. In principle a TRS on a government bond index could be disaggregated into a portfolio of IR swaps, but I think for reporting purposes it might not be.

    ReplyDelete
  15. I've spent a lot of my career investing in unfunded Total Return Swaps, and the reality is that you are not really 'swapping' a cash flow. A standard TR Swap would work like this:

    Investor goes long NDDUWI (MSCI World TR USD) for $1bn with Goldman Sachs as the counterparty.

    Goldman Sachs borrows the money at overnight rates and charges the investor a modest execution fee and finance costs of (let's say) 3m Libor + 40bps annualised. That's how GS make money.

    The investor has no counterparty risk, apart from the P&L on the trade. If the market goes up he is exposed to GS, if the market goes down GS is exposed to him and asks for margin. This exposure can be 'reset' at any point, but usually monthly.

    As the trade is unfunded, the investor has lots of cash on his balance sheet that can be invested in money market funds, short term treasuries or collateralised cash deposits. This covers some of the cost of the Libor financing.

    In effect what you really have is CFD portfolio, and I think the HP accounts should be read in this way. There is not necessarily a loser in a total return swap. Both the broker and the investor can make money.

    ReplyDelete
  16. I've spent a lot of my career investing in unfunded Total Return Swaps, and the reality is that you are not really 'swapping' a cash flow. A standard TR Swap would work like this:

    Investor goes long NDDUWI (MSCI World TR USD) for $1bn with Goldman Sachs as the counterparty.

    Goldman Sachs borrows the money at overnight rates and charges the investor a modest execution fee and finance costs of (let's say) 3m Libor + 40bps annualised. That's how GS make money.

    The investor has no counterparty risk, apart from the P&L on the trade. If the market goes up he is exposed to GS, if the market goes down GS is exposed to him and asks for margin. This exposure can be 'reset' at any point, but usually monthly.

    As the trade is unfunded, the investor has lots of cash on his balance sheet that can be invested in money market funds, short term treasuries or collateralised cash deposits. This covers some of the cost of the Libor financing.

    In effect what you really have is CFD portfolio, and I think the HP accounts should be read in this way. There is not necessarily a loser in a total return swap. Both the broker and the investor can make money.

    ReplyDelete
  17. John,

    Why is the time you've spent on this analysis a good use of that time? How does this analysis benefit your investors? Would you consider going long or short HP based wholely or partially on this analysis?

    ReplyDelete
  18. Unfortunately, we still don't know who is responsible for the swap decision. It could just as well be one of her outside managers.

    The 5500 would report the positions managed by institutional advisers as separate line items. In other words, simply looking at the pooled results of HPs investment advisers doesn't help. You have to somehow compare the derivative performance of those pooled products. Not an easy task -- particular when you can't attribute a particular derivative to a particular manager.

    Given the way ERISA liability works in the United States, in fact (i.e., the plan sponsor has a lot more liability than plan advisers) it's pretty likely that the swaps trading was done on the outside.

    I don't mean to detract from HPs performance entirely, though. They definitely did a very good job in manager selection and, likely, in asset allocation. That stuff matters too and often goes unappreciated in corporate investment management groups (particularly in non-financial companies like HP).

    ReplyDelete
  19. Has it already been pointed out to the author that Ms. Tai has only been CIO since September, 2010?

    I am also curious if the author feels that Jane Mendillo should be tarred the WORST fund manager you've never heard of, since her return in 2009 was the worst of any school, and she still has not brought Harvard's endowment up to where it was when she took over? My answer is no, because she should not be held responsible for her predecessor's decisions...but I would say that about Ms. Tai as well.

    In short, I feel Mr. Hempton is trying to construct supporting evidence for a thesis he already believes to be true.

    ReplyDelete
  20. On analysis like this?

    I told you where I started: I was trying to work out how underfunded the defined benefit scheme was.

    This is a core thing if you wish to go either long or short HP.

    The rest - well I just got curious.

    Curiousity is a clear part of the job description for a hedge fund manager I think.


    J

    ReplyDelete
  21. John, I don't understand where your readers are coming from. They should be able to figure out investment occurs after analysis. Most of the things you post warrant further attention. Had the returns been fudged, clearly this would have mattered.

    ReplyDelete
  22. John, bipolarboy has it right. The TRS are just a method of gaining additional exposure to long bonds to hedge the pension fund's interest rate risk on the DB liability, while still maintaining some exposure to equities instead of cashing out 100% to go into bonds like the Boots pension fund did in 2000-ish.

    ReplyDelete
  23. If the funded status is all you need, this is a line item in their 10-k:

    http://h30261.www3.hp.com/phoenix.zhtml?c=71087&p=irol-reportsAnnual

    Page 132: At the end of 2011 they had 10,662 in assets and 11,943 in liabilities.

    91.2% funded (92% as of FY 2010 showing AGAIN that the TR swaps were simply hedges). Not too bad. Here you can find those that were much worse (as of 2010): http://www.goldmansachs.com/our-thinking/topics/pensions-research/funded-status-takes-a-fall-doc.pdf

    ReplyDelete
  24. John- Jake is spot on, and it is such a simple answer.

    You constantly act like a Gold fish swimming round in a very small bowl and are surprised at the seemingly obvious round then next corner.

    Short,long,long, short always suspicious...so much effort dude for so little alpha.

    Go find some more gullible fools with money...

    ReplyDelete
  25. John, it's most probably TRS on the long gov/credit index or the long corporate bond index. I used to be a pension consultant, and that was a fairly common way for plans to get duration. Interest rate swaps are useful in getting duration but your discount rate is a long AA index so you need some spread exposure in there to be well hedged (plus, in 2009 that spread duration was pretty cheap). And, as one of the other folks here said, it's most likely managed by a third party manager (such as NISA, PIMCO or BlackRock).

    ReplyDelete
  26. there are some bitter Anonymous guys on here.

    ReplyDelete
  27. oh no.... what happened to Focus Media?

    Deal completed? My company block access to any china web site, so I could not tell.

    but it does look bad on bears. :(

    ReplyDelete
  28. John, you are fantastic, but this is honestly value-less without knowing what the TRS is on.

    The comments that assume they are interest rate swaps (which are separately designated) are smilarly value-less.

    My guess? TRS on long dated TIPS because that's the cheapest way to hedge long-dated liabilities right now. So she has been paying maybe OIS + 10-20 and receiving the total return on 30y TIPS. That would generate the sort of return we are looking at.

    For what it's worth, looking at just assets in pension space is also not helpful - it's vital to look at the asset-liability relationship. The liabilities are probably up 25-30% over 2 years so she needs to have done a bloody good job.

    ReplyDelete
  29. SEC still need to approve the deal, which can take 6-12 months, according to the blog of another news reporter.

    ReplyDelete
  30. I might have read this wrong, but didnt the board just accept an offer. Doesnt the buyer now have to complete its due diligence? Is the offer now unconditional?

    ReplyDelete
  31. Former pension consultant here again...

    Pete D - because pension benefits are not directly tied to inflation in the U.S. (unlike in other countries), TIPS make a lot a less sense as a liability hedge and are very rarely used in great amounts. Most plans do a combination of Treasuries (or Tsy futures or interest rate swaps) and long-dated corporate bonds (or some synthetic replication thereof). At least 3 or 4 plans that my firm consulted to bought TRS on the long credit index as part of their LDI program.

    If you want to find red flags in companies' pensions, go look for ones that have 60%+ in equity and 40% in market dur bonds (plenty of those still around despite the LDI trend). It's batshit crazy to try to defease 12-14 year duration liabilities with stocks.

    ReplyDelete
  32. "The Transaction, which is currently expected to close during the second quarter of 2013, is subject to various closing conditions, including a condition that the Merger Agreement be authorized and approved by an affirmative vote of shareholders representing two-thirds or more of the Shares present and voting in person or by proxy as a single class at a meeting of the Company's shareholders convened to consider the authorization and approval of the Merger Agreement. "

    I wonder what the other closing conditions are? I wonder why they did not see fit to specify them?

    ReplyDelete
  33. Most defined benefit funds are underfunded. They assume 8% returns in their actuarial assumptions and are two thirds are invested in debt.

    ReplyDelete