Dumb and Dumber: The Debt Valuation Adjustment Mess We're in Today
Now that DVA is back in the news thanks to Morgan Stanley's earnings report, here's what it is, why it matters, and why Morgan Stanley's earnings are more attractive than JP Morgan's.
Well, Morgan Stanley did beat Goldman Sachs and did earn more than $1 per share, and its DVA was by far its single biggest line item. Now, once again, everyone is talking about DVA and what it means. After spending some time trying to read through some FASB Statements and 10-Qs from prior quarters for JPMorgan and Morgan Stanley, I came away almost as confused as when I started. It seems fairly complex, and not surprisingly, seems short on details that would let you perform the calculations yourself. I am not, thankfully, an accountant, but I think the best hope of figuring out what these DVA numbers mean for this quarter and future quarters is worth trying to reconstruct how and why the rule exists in the first place and how it has evolved.
An Asset and a Liability
At one point in time, banks would make loans or buy mortgage bonds. They would issue debt to fund those positions. Most sat in hold to maturity books and the bank earned a net spread and only had losses if there were actual defaults. Banks were never required to "match fund" the positions, but that isn't particularly important for these purposes.
Then banks and regulators moved to mark-to-market accounting. The banks could elect to hold assets in a "trading" book and book the changes in market value into earnings each quarter. There were two reasons banks wanted to do this. One is that the world was in a bull market mode and most of the assets banks held were appreciating in value, so they wanted to be able to monetize it and book the earnings at the time of appreciation without having to wait for the net interest income to flow in over time. The second reason was that regulators agreed that an asset held in a mark-to-market book would attract less regulatory capital than the same asset held in a "hold to maturity" or "banking" book. The view was that by forcing price changes through earnings on a quarterly basis, the banks would have more focus on risk management and the regulators could be sure that nothing was hidden on the books (i.e., under-reserved), so the banks were able to grow their balance sheets.
In good times, this worked extremely well. In bad times, the higher leverage and direct hit to earnings had a major impact. It created forced liquidations, etc. The SIVs are an example of this in pure form. But before getting too far ahead of ourselves, let's look at what the bank holds now. They hold a bond in a mark-to-market account and have borrowed money to make that purchase.
At the most basic level, a good example to work with would be a bond that the bank purchased at par, and a bond they issued at par to fund that asset. If the bond they purchased is in a mark-to-market account and trades at 95 at the end of the quarter, they would have to book a 5 point loss.
There is little disagreement on whether this is the correct treatment for a reasonably liquid asset held in a mark-to-market book. Banks were happy to book gains when prices were appreciating, but this was painful in an environment where credit is performing poorly. If they just made a poor credit decision and the bond they bought is trading at 95 and the rest of the universe of similar bonds is still at par, then the banks cannot really complain. They made a bad decision.
The scenario that the banks/regulators/accountants wanted to address is: What happens if it is just part of an overall market move and all similar bonds are currently trading at 95? More specifically, what if their own debt was trading at 95? Since the bond is in a "trading" book, the bank could sell the bond it owns at 95. Then the bank could buy back its own debt at 95 and retire it. That is very normal and makes perfect sense. I don't think anyone could argue with the fact that if the bank went and bought back its own bonds at 95 that were carried at par, it would be a 5 point gain, which would offset the loss on the asset they had purchased.
But banks didn't want to have to sell bonds and buy back their debt to manage their exposure this way. Without a doubt it is cumbersome, and may encourage them to sell assets that they believe are just mispriced relative to the market, etc. Lots of good reasons to not want to sell assets and buy debts on a regular basis exist. It also means that they don't have to test the bid of the asset, they don't have to deleverage, they don't have to pay bid/offer, etc., so there are certainly some cynical reasons they wouldn't want to go through the process.
The "logical" conclusion was that banks could take a mark-to-market gain on the drop in value of their bonds. Since in theory they could buy the bonds back at current prices and retire them, they wanted to be able to book that "gain" through earnings at the same time as they book the loss on the assets.
In principle, I can understand the logic. One problem is that it doesn't force them to deleverage. It allows banks to have leverage remain high. It can relieve selling pressure on assets meaning that the system may not be fully reflecting true market value – the credit markets are illiquid relative to stocks at the best of times.
There is one other big difference between the asset and liability side. On their asset side, as the bond they bought continues to decline in price, losses continue to grow. If the asset defaults, the losses remain on the book. On the liability side, it gets very strange. What happens to a bank when it defaults? Lehman must have had a massive quarter when it defaulted because it could write down the value of the debt to $.20 on the dollar (or wherever it was trading at the time) and book a huge gain. Yes, that is where the theory really diverges from reality.
If a bank sells an asset and buys its debt back, it has reduced leverage. If the bank were ever to default, there are fewer assets, but the debt level is lower. That should result in higher recoveries for debtholders, and make it easier to raise equity capital as bankruptcy becomes a real possibility.
With the DVA accounting, the situation is different. If the bank's bonds are trading at 60, then the bank will have booked a gain of 40 on those bonds, but the claim of the bondholders is still 100. They write down the value of the bonds to where they could buy them, but the claim on the company has not changed -- whereas if they had purchased the bonds back, the total claim by debtholders would have been reduced. You could argue that the bonds could be bought back post bankruptcy, but as soon as the bank has entered bankruptcy, everything changes. It is not fluid. It is a step function. The bonds in the real world may rally, but this accounting methodology shows its weakness – allowing leverage to exist and never testing whether the bid side of the assets is real and if anything could truly be liquidated at the mark.
DVA Matched Books
In general, I'm not a huge fan of the rule, but I can see that if it was strictly controlled, it would make some sense. If there was a "DVA" book where assets and liabilities were kept and specifically designated as being in the "DVA" book, then I would be a lot more comfortable. Then you could see which assets decreased in value and which liabilities decreased in value (causing gains). That level of transparency wouldn't solve the issues of letting leverage remain too high in a down market and not really challenging the bids of the assets, but at least then someone could analyze the positions, and get comfortable with the accounting.
This book would effectively be a pure "alpha" book. Gains or losses in any quarter would reflect the credit picking skills of the bank. If the assets outperformed their liabilities (i.e., they made good decisions), the book would have a gain. In any quarter where their liabilities outperformed their assets, they would have a loss, and investors could question why they were so bad at making credit decisions. But creating something relatively transparent to reduce the inherent complexities and risks is just not how the banks work.
The DVA mess we have today
Here is all that I can figure out so far:
The debt outstanding is using the DDIS function on Bloomberg. The assets are based on year-end balance sheets. The DVA for JPMorgan and Morgan Stanley are both pre-tax numbers. The C number is actually a CVA (I will admit to not knowing the difference yet), and I have not been able to tell if it is pre- or post-tax. The CDS change is for 5 year CDS.
If the DVA is meant to offset losses on the asset side, either Morgan Stanley had a far worse portfolio (which I doubt) or over time the ideal of DVA has been lost in the massive pile of line items on bank books. The DVA seems more correlated to CDS spread move than anything else. I am trying to find more about what is actually the difference between DVA and CVA, but on the surface it would seem that Morgan Stanley either had really awful assets that will bounce back hard with their credit, or a lot of their DVA will go away next quarter (if their spreads remain normalized between now and year end).
One thing I liked about JPMorgan's earnings call and their potential next quarter to offset a DVA reversal is they seemed to be conservative on releasing loan loss reserves and seemed to add significantly to some other reserves. Jamie Dimon talked about how DVA earnings were low quality. My guess is that they are prepared and if the DVA reverses next quarter, they will have asset gains and reserve releases against them.
Compare that to James Gorman, who I don't think "gets it". He seems to be touting the strength of earnings based on DVA. Nothing about his actions shows that he expects them to reverse next quarter, but on the other hand, there is nothing I have seen or heard so far that indicates his assets will massively outperform those of his peers next quarter. Goldman Sachs seemed to have neither CVA nor DVA.
Any help to clarify the CVA and DVA numbers would be appreciated, but so far I would have to say that JPMorgan's fourth quarter will depend on its fourth quarter performance and shouldn't be impacted much by a reversal in DVA if things go well. For Morgan Stanley, I would be very careful as the CEO seems to happy with the number, and I would rather believe that their credit spreads massively underperformed the market, than that the assets they own did so horribly they can bounce back even to offset the DVA reversal.
Since deferred comp is a big deal for banks and they have to constantly accrue for prior year bonuses, the worse a stock price does in a quarter, the less of a charge the bank has to take for prior year earnings. Not only is the amount they planned on accounting for that period lower (since it is tied to stock price), but they may have over-reserved in prior quarters when they took reserves against higher stock prices. Again, this would make Morgan Stanley's earnings more attractive than JPMorgan's.
Editor's Note: For more from Peter Tchir, check out TF Market Advisors.
Copyright 2011 Minyanville Media, Inc. All Rights Reserved.
Daily Recap Newsletter