What is the Impact of Derivatives on US Banks?

[LaurieAG]

The Office of the US Currency Comptroller has an interesting chart 4 titled '5 Banks Dominate in Derivatives, Insured U.S. Commercial Banks, Second Quarter 2011'.

http://www.occ.gov/topics/capital-markets/financial-markets/trading/derivatives/dq211.pdf

I'm speechless!

 

[John Creighto]

The Office of the US Currency Comptroller has an interesting chart 4 titled '5 Banks Dominate in Derivatives, Insured U.S. Commercial Banks, Second Quarter 2011'.

http://www.occ.gov/topics/capital-markets/financial-markets/trading/derivatives/dq211.pdf

I'm speechless!

Reading the executive summary the only line that concerns me is:

"The notional amount of derivatives held by insured U.S. commercial banks increased $5.3 trillion, or 2.2%, from the first quarter of 2011 to $249 trillion. Notional derivatives are 11.6% higher than a year ago."

I'll read the article in full when I get home.

What concerns me most about this statement is how these assets will show up in the balance sheets of banks and whether any part of these fictitious assets will show up in the calculation of their capital for the purposes of their capital adequacy requirements. If leverage is allowed based capital comprised of derivative assets (I’m not sure if it is), then this means that banks can expand the money supply based largely on assets that have no relation to any physical capital. Well, some money is required for liquidity for trading these derivative assets, I don’t believe it should directly impact the money supply in the real economy. Derivative contracts should not get any precedence in the event of a bankruptcy. That is when there isn’t enough real money/capital to cover the loans/bets, lending based on real assets should take precedence.

 

[marcus]

Graph 5A gives the top five in terms of involvement with derivatives:

Bank of America
Citibank
GoldmanSachs Bank
JPMorgan Chase Bank
HSBC Bank

Graph 4 shows their involvement is overwhelmingly more than all the other banks put together.

The heading on Graph 5A is "Percentage of Total Credit Exposure to Risk Based Capital
Top 5 Insured U.S. Commercial Banks by Derivative Holdings"

It shows how each of the bank's credit exposure to stuff like derivs. has changed from year to year. HSBC shows a gradual decline. the others not.

Thanks for finding this. And thank you both for commenting!

 

[BWV]

you have to take numbers like $249 trillion (over 4x global GDP) with a grain of salt

the calculations generally do not net out offsetting exposures, which is much of what banks do when they underwrite derivatives

For example, they might offer one client an currency swap where bank receives Euros and pays dollars and then offer another client the opposite swap - paying Euros and receiving dollars. The bank makes a small spread on the notional value and the risks are offset, but the full notional values of both swaps show up in these derivative numbers.

 

[John Creighto]

you have to take numbers like $249 trillion (over 4x global GDP) with a grain of salt

the calculations generally do not net out offsetting exposures, which is much of what banks do when they underwrite derivatives

For example, they might offer one client an currency swap where bank receives Euros and pays dollars and then offer another client the opposite swap - paying Euros and receiving dollars. The bank makes a small spread on the notional value and the risks are offset, but the full notional values of both swaps show up in these derivative numbers.

If they are offering clients swaps they are selling futures not holding derivatives as assets. Additionally, in the case you mentioned until the contract expires there is still a liability (assuming the bank is the counter party ) because only one party is going to exercise the swap.

 

[WhoWee]

I'm speechless!

I consider that a healthy response.

my bold
"Notionals

Changes in notional volumes are generally reasonable reflections of business activity, and therefore can provide insight into potential revenue and operational issues. However, the notional amount of derivatives contracts does not provide a useful measure of either market or credit risks.

The notional amount of derivatives contracts held by insured U.S. commercial banks in the second quarter increased by $5.3 trillion (2.2%) to $249 trillion from first quarter 2011. The notional amount of derivatives is 11.6% higher than a year ago.

The five banks with the most derivatives activity hold 96% of all derivatives, while the largest 25 banks account for nearly 100% of all contracts."

I take very little comfort in knowing the top 5 banks hold 96% of the contracts and that activity is up 11.6% - that (IMO) infers they might be scrambling to balance their portfolios.

 

[BWV]

If they are offering clients swaps they are selling futures not holding derivatives as assets. Additionally, in the case you mentioned until the contract expires there is still a liability (assuming the bank is the counter party ) because only one party is going to exercise the swap.

I don't understand the comment - futures are derivatives and swaps are not futures (although you can replicate vanilla swap economics with futures). Furthermore there is no "exercising" a swap - it is not an option. A swap is active the moment the contract is signed and there are penalties to exiting it prematurely.

 

[John Creighto]

I don't understand the comment - futures are derivatives and swaps are not futures (although you can replicate vanilla swap economics with futures). Furthermore there is no "exercising" a swap - it is not an option. A swap is active the moment the contract is signed and there are penalties to exiting it prematurely.

Oh. I thought they were like options. My bad.

 

[WhoWee]

Oh. I thought they were like options. My bad.

When ever the topic of derivatives comes up - I like to remind everyone the most successful investor in the history of the modern world (Warren Buffet) explained in a letter to his shareholders a few years ago that his choice (basically) was not to participate.

 

[John Creighto]

Aside from the discussion about derivatives there is an interesting discussion in the document about risk management using VaR (Value at risk). I am unclear what assumptions are made about the statistics and sample space, in trying to inductively determine the risks.

I do know the statistics of financial events are tail heavy but the central limit theorem tells us that even tail heavy statistics should average out eventually to Gaussian statistics provided the statistics are ergodic stationary and bounded. However, there is no limit to the number of measurements required to get a degree of confidence in the estimate of the statistics.

Var's measure the maximum risk one is likely to see for a given confidence interval. This is from the article:

VaR. For example, JP Morgan, Goldman Sachs and Morgan Stanley calculate VaR using a 95% confidence interval. If those firms used a 99% confidence interval, as does Bank of America and Citigroup, their VaR estimates would be meaningfully higher. The data series used to measure risk also is an important factor in the calculated risk measure. Firms using a longer period over which to measure risk may include the higher volatility period of the financial crisis, and therefore their measured VaR will be higher than firms that use a less volatile data series. Indeed, one major reason for the decline in VaR at large trading firms is the lower volatility environment that has prevailed since the end of the financial crisis. The VaR measure for a single portfolio of exposures will be different if the time period used to measure risk is not the same. To test the effectiveness of VaR measurement systems, trading institutions track the number of times that daily losses exceed VaR estimates. Under the Market Risk Rule that establishes regulatory capital requirements for U.S. commercial banks with significant trading activities, a bank?s capital requirement for market risk is based on its VaR measured at a 99% confidence level and assuming a 10-day holding period.Banks back-test their VaR measure by comparing the actual daily profit or loss to the VaR measure. The results of the back-test determine the size of the multiplier applied to the VaR measure in the risk-based capital calculation. The multiplier adds a safety factor to the capital requirements. An "exception" occurs when a dealer has a daily loss in excess of its VaR measure. Some banks disclose the number of such "exceptions" in their published financial reports. Because of the unusually high market volatility and large write-downs in CDOs during the financial crisis, as well as poor market liquidity, a number of banks experienced back-test exceptions and therefore an increase in their capital multiplier.

This suggests that most firms are computing their VaR purely based on the historical data of the firm. Well this may be okay since typically a VaRs predict the expected maximum loss within a confidence interval over a short period of time (one to two days typically but 10 days for capital requirement risk assessment) it is only valid if the recent past statistics of the firm give a good measure of future performance of the firm.

This is likely not the case in an irrational market and as a consequence sampling should be done over a long enough period of time to account for short term market irrationality. Well, over a long period of time the volatility of a firms stock could change significantly one would hope that there are metrics of a firms performance (such as price to earnings and leverage) whose statistics better corrolate too loss risk over longer intervals of time.

Additionally, in the grand scheme of things 95% confidence bounds on daily risk may not be relevant because in two days our confidence of an not seeing an event that falls outside of these bounds drops to 90.25 % and after about 14 days we can't even be 50% confident. In other words 95% confidence says, in one day we are probably safe but in about two weeks it's a crap shoot. Of course, I'm presuming successive days as independent events if but if there is a lot of low frequency noise (auto regressive) the confidence won't fall off so quickly.

Finally pay attention to the parts I bolded. Banks are risk weighting their capital based on purely inductive statistics but do back test their models and add a multiplier to try to give a safety factor for errors. They use a 99% confidence level and a 10 day holding period for their capital requirements. I would like to dive into the statistics of this further but perhaps in another thread. But before I do, does anyone beside me think it's odd that financial firms are reporting their Var's with less robust confidence levels and time frames then is required to meet capital requirement regulations.

 

[WhoWee]

Aside from the discussion about derivatives there is an interesting discussion in the document about risk management using VaR (Value at risk). I am unclear what assumptions are made about the statistics and sample space, in trying to inductively determine the risks.

I do know the statistics of financial events are tail heavy but the central limit theorem tells us that even tail heavy statistics should average out eventually to Gaussian statistics provided the statistics are ergodic stationary and bounded. However, there is no limit to the number of measurements required to get a degree of confidence in the estimate of the statistics.

Var's measure the maximum risk one is likely to see for a given confidence interval. This is from the article:



This suggests that most firms are computing their VaR purely based on the historical data of the firm. Well this may be okay since typically a VaRs predict the expected maximum loss within a confidence interval over a short period of time (one to two days typically but 10 days for capital requirement risk assessment) it is only valid if the recent past statistics of the firm give a good measure of future performance of the firm.

This is likely not the case in an irrational market and as a consequence sampling should be done over a long enough period of time to account for short term market irrationality. Well, over a long period of time the volatility of a firms stock could change significantly one would hope that there are metrics of a firms performance (such as price to earnings and leverage) whose statistics better corrolate too loss risk over longer intervals of time.

Additionally, in the grand scheme of things 95% confidence bounds on daily risk may not be relevant because in two days our confidence of an not seeing an event that falls outside of these bounds drops to 90.25 % and after about 14 days we can't even be 50% confident. In other words 95% confidence says, in one day we are probably safe but in about two weeks it's a crap shoot. Of course, I'm presuming successive days as independent events if but if there is a lot of low frequency noise (auto regressive) the confidence won't fall off so quickly.

Finally pay attention to the parts I bolded. Banks are risk weighting their capital based on purely inductive statistics but do back test their models and add a multiplier to try to give a safety factor for errors. They use a 99% confidence level and a 10 day holding period for their capital requirements. I would like to dive into the statistics of this further but perhaps in another thread. But before I do, does anyone beside me think it's odd that financial firms are reporting their Var's with less robust confidence levels and time frames then is required to meet capital requirement regulations.

IMO - increased activity = risk mitigation.

 

[LaurieAG]

you have to take numbers like $249 trillion (over 4x global GDP) with a grain of salt.

Yeah, the total risk is only between 2 and 8 trillion dollars and the money is insured.

 

[WhoWee]

Yeah, the total risk is only between 2 and 8 trillion dollars and the money is insured. Great.

DEEP breaths Laurie AG - deep breaths!

 

[LaurieAG]

They use a 99% confidence level and a 10 day holding period for their capital requirements.

That reminds me of the ancient Greek roots of 95% confidence. After 19 years absence, the suitors and their girlfriends had 95% confidence that Odysseus would not return from the Trojan wars within 20 years.

 

[WhoWee]

That reminds me of the ancient Greek roots of 95% confidence. After 19 years absence, the suitors and their girlfriends had 95% confidence that Odysseus would not return from the Trojan wars within 20 years.

Look at it this way Laurie - as long as the rules of the game don't change- the players will tweak their positions just enough that nobody will understand their condition.

 

[John Creighto]

Reading the executive summary the only line that concerns me is:

"The notional amount of derivatives held by insured U.S. commercial banks increased $5.3 trillion, or 2.2%, from the first quarter of 2011 to $249 trillion. Notional derivatives are 11.6% higher than a year ago."

I wanted to put this in perspective. The total value of all wealth in the United States is 55 trillion:
https://www.physicsforums.com/showpost.php?p=3548782&postcount=23

So the notional amount of derivatives held by U.S. commercial banks is five times the total wealth of the united states.

Now with regards to my previous comment about swaps. Even though swaps aren't like options so if you hold two opposite positions some of the risk should cancel, there is a possibility that one of the parties can't pay so there is some counter party risk and consequently the bank should be required (if they aren't already) to represent some of this as a liability on their balance sheets and accurately weight the risk when doing capital adequacy calculations.

 

[LaurieAG]

I wanted to put this in perspective. The total value of all wealth in the United States is 55 trillion:

This is from Money Morning, who provided the original link.

http://moneymorning.com/2011/10/12/derivatives-the-600-trillion-time-bomb-thats-set-to-explode/

The world's gross domestic product (GDP) is only about $65 trillion, or roughly 10.83% of the worldwide value of the global derivatives market, according to The Economist.
...
To be fair, the Bank for International Settlements (BIS) estimated the net notional value of uncollateralized derivatives risks is between $2 trillion and $8 trillion, which is still a staggering amount of money and well beyond the billions being talked about in Europe.

Because the US government has not required banks to separate their investment arms it will ultimately pay the price for these risks as the insurer if they go bad.

 

[mheslep]

you have to take numbers like $249 trillion (over 4x global GDP) with a grain of salt

the calculations generally do not net out offsetting exposures, which is much of what banks do when they underwrite derivatives

For example, they might offer one client an currency swap where bank receives Euros and pays dollars and then offer another client the opposite swap - paying Euros and receiving dollars. The bank makes a small spread on the notional value and the risks are offset, but the full notional values of both swaps show up in these derivative numbers.

Exactly. So in that sense these figures are more akin to stock market volume than asset value. For instance the annual dollar volume of the NYSE, just one exchange, is on the order of $10 trillion, the NASDAQ another $15 trillion, etc. Trading volume in the stock market is not the same as asset value either.

 

[mheslep]

I wanted to put this in perspective. The total value of all wealth in the United States is 55 trillion:
https://www.physicsforums.com/showpost.php?p=3548782&postcount=23

So the notional amount of derivatives held by U.S. commercial banks is five times the total wealth of the united states.

That takes the derivatives discussion out of perspective, comparing apples and oranges. The dollar value of those contracts amassed over time by those banks do not equate to the same dollar amount of wealth, or assets, or liabilities, and certainly not 5X that of the US's wealth or anything like it.

Good description of the difference between the two here:
https://www.physicsforums.com/showpost.php?p=3551926&postcount=256

 

[LaurieAG]

That takes the derivatives discussion out of perspective, comparing apples and oranges.

the Bank for International Settlements (BIS) estimated the net notional value of uncollateralized derivatives risks is between $2 trillion and $8 trillion

First CERN and now BIS, do you want to let them know of their error?

 

[WhoWee]

First CERN and now BIS, do you want to let them know of their error?

This might help clarify a bit?
http://www.bis.org/publ/otc_hy1105.htm

"After contracting by 4% in the first half of 2010, total notional amounts outstanding of over-the-counter (OTC) derivatives rose by 3% in the second half, reaching $601 trillion by the end of December 2010.
Notional amounts outstanding of credit default swaps (CDS) continued to contract, falling by 1% after the 7% decline in the first half. Gross market values of all OTC contracts fell by 14%, driven mainly by the 17% decline in the market value of interest rate contracts. CDS market values shrank by 19%.
Overall gross credit exposure dropped by 7% to $3.3 trillion, compared with a 2% increase in the first half of 2010."

full text
http://www.bis.org/publ/otc_hy1105.pdf

 

[mheslep]

First CERN and now BIS, do you want to let them know of their error?

You misread. I did not say or imply the dollar value of all derivative contracts somehow netted out to zero, only that that they are not dollar for dollar in the same category as used when totaling US wealth. The BIS figure implies the same thing.

 

[LaurieAG]

You misread. I did not say or imply the dollar value of all derivative contracts somehow netted out to zero, only that that they are not dollar for dollar in the same category as used when totaling US wealth. The BIS figure implies the same thing.

My apologies, I'm going to have to put whenever I'm being sarcastic.

How about calculating the LTCM exposure vs its loss to get a rough figure for a loss based on the current magnitude. The child of LTCM used less leverage but also used the same basic principles and recently had to fold after losing 44% of one fund.

From the LTCM wiki.

It had off-balance sheet derivative positions with a notional value of approximately $1.25 trillion, most of which were in interest rate derivatives such as interest rate swaps.
...
The total losses were found to be $4.6 billion.
...
After helping unwind LTCM, Meriwether launched JWM Partners. Haghani, Hilibrand, Leahy, and Rosenfeld all signed up as principals of the new firm. By December 1999, they had raised $250 million for a fund that would continue many of LTCM's strategies—this time, using less leverage.[31] With the Credit Crisis, JWM Partners LLC was hit with 44% loss from September 2007 to February 2009 in its Relative Value Opportunity II fund. As such, JWM Hedge Fund was shut down in July 2009

.

 

[LaurieAG]

This might help clarify a bit?
http://www.bis.org/publ/otc_hy1105.htm

Thanks WhoWee, that's a good link.

The total notional amounts outstanding / gross market value = 28.4.

A gross market value of 21 Trillion is still huge.

 

[BWV]

You misread. I did not say or imply the dollar value of all derivative contracts somehow netted out to zero, only that that they are not dollar for dollar in the same category as used when totaling US wealth. The BIS figure implies the same thing.

to be clear, the dollar value of all derivative contracts in the global economy does net to zero by definition

- per the BIS it may not for a particular segment of the economy, such as US banks

 

[mheslep]

to be clear, the dollar value of all derivative contracts in the global economy does net to zero by definition

- per the BIS it may not for a particular segment of the economy, such as US banks

Yes point taken.