« I’d Stay and Be a Tourist but Can’t Take the Gun Play | Main | A Big Day in Bagram »

April 02, 2009


I don't understand why everyone can't opt in to a mutual indemnification to accept 3rd-party valuations of their securitized assets. That would solve this whole problem.

I just feel that any institution that wishes to use this accounting model must be corporately located within the US for accounting and tax purposes.

If I had to sell my house tonight, it would be sold for far less than it is worth in a rational market, where I could market it for several months, and it would be worth more.

I don't think my house is only worth the instantaneous value I can get for it, and I don't think my lenders should tell me that the house is only worth as much as I can get for it in the next 24 hours.

I think, therefore, that the value of something as expensive and confusing as these types of assets should not be valued in a 90 day window: it is not an accurate assessment of what the bank would actually sell it for.

James Kwak's afterthought at The Baseline Scenario:

Update: Here’s a thought. What if the function of these rule changes is to make it easier for banks to ignore the results of the PPIP auctions? For example, Bank A puts up a pool of loans for auction, but doesn’t like the winning bid and rejects it; Bank A doesn’t want to be forced to write down its loans to the amount of the winning bid. Or, alternatively, Bank B sells a security to a buyer, and Bank A holds the same security; Bank A doesn’t want to be forced to write down the security to the price of Bank B’s transaction.

The change to fair value accounting (Rule 157) may make it easier to claim that the sale by Bank B was a “distressed sale,” meaning it can ignore it for valuation purposes. Even if it can’t ignore the sale, the change to other-than-temporary impairment may make it easier for Bank A to classify any impairment as temporary and therefore avoid an income statement hit. You’d have to be a specialist to know the answers for sure, but in any case these rule changes don’t make it any harder.

jrudkis: the question is: what is an accurate value? Pretty much no one answers: the amount you could get for your house if you had to list and sell it tonight. Instead, people rely on comparable recent sales -- sales that did not have to happen within 24 hours, but involved the usual listing, people seeing the house over some weeks, bargaining, agreeing on a price, etc.

That seems fair -- or at least, better than any other method of valuing a house I can think of. Certainly better than letting you come up with your own model for valuing your house.

The FASB isn't really independent. It never has been. The responsibility for setting accounting standards is clearly given to the SEC under statutory law. Most of the time, it chooses to delegate that authority to a private entity, the APB until 1973, and the FASB since. Not always, though. The SEC has made declarations of what it considers to be acceptable accounting standards, and these take precedence if they conflict with pronouncements of the FASB. The idea of preserving its independence is a canard; this is a decision that is supposed to be made by a governmental organization, so pressure is perfectly appropriate.

I'm also not convinced that this is nearly as big a deal as is being assumed. For one thing, the basis of this rule was already in FAS 157. This loosens it some, but it isn't true that this was not already a factor. It was already the case that a sale in bankruptcy was not considered a factor in determining that there is an active market.

The problem isn't that this allows companies to use an inaccurate price. It's that there really isn't an accurate price for a lot of this stuff. FAS 157-e can be abused, but it is also basically correct. In the situations it describes, you can't get a fair market price. It doesn't exist.

This actually goes to one of my problems with accounting as it is currently practiced. Accounting values precision above all things. When you are putting together financial statements, you have to arrive at a value for everything. In accounting for contingencies, such as a pending lawsuit, if you can't give a reasonable estimate for what the damages will be, you omit it entirely, giving it a value of $0, and just disclose it in the notes. With complicated assets like CDOs, you have to come up with a single number to value it at.

Clearly, this isn't correct. It's valuing precision over accuracy. The real value of the asset is something between X and Y. One of the funny things about FAS 157 is that it brings statistical methods into the procedure for determining what something is worth, and then ignores the most important rule of statistics, which is that there isn't a single answer. Everything in statistics is a probability distribution over a range. FAS 157 ignores this.

I really think that accounting needs an overhaul, to figure out a way to really apply this. There are a lot of assets and liabilities for which it is misleading to give a single value. We need a way to represent this. Now, I'm just a guy with an undergraduate degree in accounting studying for the CPA exams, so I have neither the breadth of knowledge nor the time to try to come up with the solution to this problem, but there should be some professors somewhere thinking about this.

The last note is that people complaining about this should, in fact, like the Geithner plan for public/private auctions of bad assets. Read FAS 157-e, and then go back and read white paper on the program. Note that the auctions proposed will meet the requirements for producing an orderly market. FAS 157-e won't apply to anything that is comparable to an asset that gets auctioned.

Baloney. People really need to read the proposal. In order to declare that prices are produced by an inactive market, you need to go through a two step process. The first step is where the question about bankruptcy or forced sales come in. I agree that the PPIP auctions will be conducted in a way that suggest that there is an inactive market under Step 1.

However, you still have to go through Step 2:

13. If the reporting entity concludes in step 1 that the market for the asset is not active, then the reporting entity will proceed to step 2. In step 2, the reporting entity must presume that a quoted price is associated with a distressed transaction unless the reporting entity has evidence that (a) there was sufficient time before the measurement date to allow for usual and customary marketing activities for the asset and (b) there were multiple bidders for the asset.

It is very clear that the auctions will meet both of these criteria. They will not produce prices in an inactive market.

The other proposal he mentions, dealing with other-then-temporary impairments, is here. It just doesn't make much difference. It makes the definition of the assertion that must be made to declare an impairment temporary more specific, but not substantially different. It allows a company to record an other-than-temporary impairment of a debt instrument for credit reasons (as opposed to, say, interest rate changes) in Other Comprehensive Income rather than Net Income. It must amortize the impairment into Net Income over the life of the instrument. It must also reflect the impairment on the balance sheet immediately, so it's not like it will be hidden. You just have to read more than the income statement, which you should do anyway. Third, it allows that equity assets can be temporarily impaired, but that has no relevance to the PPIP auctions.

My "Baloney" is to James Kwak's claim. The system is cutting off initial quotes that I make in my posts.

The problem isn't mark-to-market per se. Warren Buffet thinks that you should keep mark to market for disclosure purposes (to keep track of the changing health of the company) but that you should use something like a 5 year moving average of market values for regulatory capital purposes. The problem with mark to market for regulatory capital purposes is that it will force you to sell directly into the fire sale to raise cash, depressing the price further and causing the problems to spread.

Yes, assets should be priced according to the market, but a rolling average would be much less pro-cyclical, and makes sense for assets you have some choice about when to sell.

You are saying the assets are valued on the basis of the value of yield it makes but the assets will be valued as the market value.SEO Tampo

Relaxing mark to market certainly is consistent with the Geithner plan and the Obama administration's broad views on mortgage-backed securities. If those assets actually turn out to be worth exactly what the market says they're worth right now... we have deeper and more fundamental problems than the precise rules of bank accounting.

I thought one of the big selling points of free markets is efficient price discovery. What's the difference, really, between mark-to-model and any kind of planned economy, or someone designates a price for a commodity/asset?


//jrudkis: the question is: what is an accurate value? Pretty much no one answers: the amount you could get for your house if you had to list and sell it tonight. Instead, people rely on comparable recent sales -- sales that did not have to happen within 24 hours, but involved the usual listing, people seeing the house over some weeks, bargaining, agreeing on a price, etc.//

In the current environment it is just not true that recent sales are not fire sales. In some markets, maybe a third or half of sales are foreclosures. Foreclosures, are auctioned on the courthouse steps. The buyer has to pay the full price immediately with a cashiers check and there is no ability to inspect the house beforehand. Further, there is a risk that the buyer may have to evict someone from the home. This results in an extreme discount. In a more orderly market, where there is time for inspections and time to get mortgage financing the prices are more normal. I know this because I have attended many of these auctions. My son bought a house at auction in August for $237k and resold it in December in the more normal way for $329k. He then bought another at auction in December for $248k and sold it two weeks ago for $365k (escrow has not closed yet.)

Mark to market is forcing banks to unload these houses in fire sales instead of in a more orderly way. The only people who profit from this are people who can buy and are willing to buy houses for cash on a moment's notice. There are very few people like that.

*** Here come the critics who'll say I am a liar ***

hilzoy and Sebastian seem to agree that mark-to-market should be kept for disclosure purposes, but that regulations might need to be changed to allow for some problems this creates.

Count me in.

I wonder what everybody (bankers, financiers) thought about mark-to-market when the market was on its way up.

In any case, mark to market for disclosure, rolling average for regulatory purposes seems reasonable.

"I wonder what everybody (bankers, financiers) thought about mark-to-market when the market was on its way up."

They loved it because it is somewhat self amplifying in whatever direction the market is travelling. But it is actually more of a serious feedback loop on the way down. This isn’t immediately intuitive so I’ll try to explain: On the way up a bubble, mark-to-market is self reinforcing because it inflates the value of assets beyond their ‘normal’ expected return. As the assets look better, more people end up trying to acquire them, which pushes the price up further via supply/demand. However, this is occurring in a liquid and expanding market, so while it is self-reinforcing it doesn’t immediately jump upward in value.

In an illiquid and contracting market, the self-reinforcing nature of mark to market is especially harsh. Buying a house is something that you can put off a few months while things calm down. This alone (but especially combined with overbuilding from the bubble) can create a situation where there are many more buyers than sellers, so that the house won’t sell. This makes the mark-to-market price effectively zero or near zero. The long term worth of the house isn’t zero or near zero (in the sense that if you hold on to it, it will almost certainly be worth some tens of thousands of dollars at a very minimum even in heavily depressed areas). But because mark-to-market functions best in a liquid market, it overshoots the lower boundary of the value all the way to near-zero in an illiquid market.

Now in a world without capital requirements that wouldn’t be a big deal. The bank or whatever would just hold on to it. The problem is that capital requirements mandate that you hold a certain value of a certain quality of asset. Mark to market lowers the value AND quality of the asset in an illiquid environment. So you must then replace it with cash or some other higher quality investment. The problem there is that it pretty much forces a sale of the illiquid assets directly into the bad environment, which drives the price down even more quickly while devaluing similar assets for other people, spreading the problem further.

As for the idea that prices are the best discovery mechanism, they actually seem to be. But they aren’t a PERFECT discovery mechanism. And they are much less reliable than usual in an illiquid environment, especially in an environment where regulatory rules are forcing sales from people that wouldn’t normally want or have to sell right this moment.

I suggested that mark to market use average values six months ago here. But I still think that a static rolling average won't be sufficient: it will be very hard to choose a single static value for how much weight your rolling average gives older data that both prevents problems during bubble inflation and deflation while avoiding undue inhibition of price transmission during non-bubble times. The problem seems analogous to control problems for which the Kalman Filter is applied: you need to constantly integrate two sources of data together but there is no single static weight that is always optimal for doing so, so you have to calculate that weight dynamically.

If not for the issuance of FAS157 effective November 15,2007 by the SEC and FASB . This rule requires banks to mark to market instead of mark to model as previously calculated since the inception of FASB. Pre FAS 157 when an investment vehicle trades thinly (such as mortgage backed securities) the underlying holder of the investment is allowed to use its own assumptions on fair value, today it must be valued based on what it can be sold at in the open market. If I'm a bank and expecting to receive future cash flow of $1000 a month over the next 30 years on an loan of 200K for a total payment of 360k, based on todays fair market value and panic in the market the underlying issue has been written down to 40 cents on the dollar or approximately 80k and this is a conservative figure.

In the current market it would not be wise to sell any of the troubled assets based on fair market valuation as there are not enough buyers which drastically reduces the price of the asset. This rule change with the mark to market update should slowly begin to bring back some credibility to this market.

The comments to this entry are closed.

Blog powered by Typepad