My friend Justin sent me this article
Bloomberg: Naked Ban Means CDS Safest Relative to Germany: Poland Credit
"The extra cost of insuring Polish government bonds for five years with
default swaps, or CDS, over Germany fell to 45 basis points on Oct. 11,
the smallest gap since August 2008, a month before Lehman’s crash
deepened the global financial crisis. The yield on benchmark 10-year notes fell to a seven-year low of 4.46
percent yesterday, according to data compiled by Bloomberg. European Union regulations will from Nov. 1 ban ownership of sovereign
credit-default swaps without holding the underlying debt assets, known
as naked long positions, to prevent investors from pushing up insurance
costs. Bond holders are lured to Poland by its economic growth, budget
discipline and speculation the central bank will lower interest rates. "
Most of the article is about the relative credit-worthiness of Poland. However the part that I found most interesting is the part about the new EU rule. Here is a summary of the EU rule
Harvard Law School Forum: Europe Restricts “Naked” Credit Default Swaps and Short Sales
"On November 15, 2011, the European Parliament adopted a regulation banning any person or legal entity in the European
Union (“EU entities”) from entering into “naked,” or uncovered, credit
default swaps (“CDS”) on sovereign debt and restricting uncovered short
sales on shares and sovereign debt (the “Regulation”) after November 1,
2012. [1]
The Regulation also bans such transactions from being effected on any
trading venue in the European Union (the “EU”). CDS on sovereign debt
that do not hedge exposure to the sovereign debt itself or to assets or
liabilities whose value is correlated to the value of the sovereign debt
will no longer be permitted. Short sales of shares and short sales of sovereign debt will be
permitted only where the seller has “located” the share or debt
instrument prior to entering into the agreement and has a “reasonable
expectation” of being able to borrow the shares. The Regulation provides
exemptions for market making activities and primary market operations
and allows Member States of the EU (“Member States”) to temporarily
suspend the ban on uncovered CDS on sovereign debt if the Member State
determines that its sovereign debt market is not functioning properly as
a result of the ban. The Regulation also introduces reporting
requirements for significant net short positions"
So the EU is not banning short sales of sovereign debt - it just requires that you first locate the bond that you intend to short - like we are required to do in the US equity markets. If the EU did not also restrict holding uncovered CDS then speculators could get around the locate rule by synthetically generating the same exposure in CDS. So why would the EU want to do this? From Harvard Law School Forum
"The European Commission (the “Commission”) proposed the ban on uncovered
CDS and limitations on uncovered short sales in response to recent
market volatility in Euro-denominated sovereign bonds. Member States
reacted differently to the market volatility, some taking no action at
all. The Commission proposed the Regulation to reduce regulatory
arbitrage and compliance costs arising from a fragmented regulatory
framework across the EU. While the Commission acknowledges
that short sales and CDS have economic benefits, they also note that
these transactions present risks to the market that must be controlled. [2]"
An (unspoken) added benefit may be the following: by banning naked long CDS on sovereign bonds the rule should decrease the demand for CDS protection on such bonds. Decreasing the demand for CDS protection on sovereign bonds "may" reduce the required insurance premium for these bonds. Since investors will have to pay less to insure their sovereigns that could reduce the required yield on the bonds (maybe) and help the issuing countries.
Sunday, October 28, 2012
Interesting Hypothesis on the Red Blue Divide
NYT: Why Are States So Red and Blue - Stephen Pinker
"The historian David Hackett Fischer traces the divide back to the British settlers of colonial America. The North was largely settled by English farmers, the inland South by Scots-Irish herders. Anthropologists have long noted that societies that herd livestock in rugged terrain tend to develop a “culture of honor.” Since their wealth has feet and can be stolen in an eye blink, they are forced to deter rustlers by cultivating a hair-trigger for violent retaliation against any trespass or insult that probes their resolve. Farmers can afford to be less belligerent because it is harder to steal their land out from under them, particularly in territories within the reach of law enforcement. As the settlers moved westward, they took their respective cultures with them."
"The historian David Hackett Fischer traces the divide back to the British settlers of colonial America. The North was largely settled by English farmers, the inland South by Scots-Irish herders. Anthropologists have long noted that societies that herd livestock in rugged terrain tend to develop a “culture of honor.” Since their wealth has feet and can be stolen in an eye blink, they are forced to deter rustlers by cultivating a hair-trigger for violent retaliation against any trespass or insult that probes their resolve. Farmers can afford to be less belligerent because it is harder to steal their land out from under them, particularly in territories within the reach of law enforcement. As the settlers moved westward, they took their respective cultures with them."
Saturday, October 27, 2012
Wojtek the Bear
Wojtek the Bear by Aileen Orr.
It is a great story. In 1942 the 22nd Company, Polish Army Service Corps (Artillery) while marching through Persia adopted a Syrian brown bear whom they named Wojtek (pronounced Voy-chek). Over the next three years as they traveled through the Mideast to Italy and eventually to Scotland Wojtek traveled with them. (see here for their travels.) Knowing no other life Wojtek adopted astonishingly domestic and nearly human behavior. It was frequently said that he thought of himself as one of the company. He lived in tents with them, drank beer and smoked cigarettes with them, wrestled, and responded to their Polish language. In order to get him rations the Poles registered him as Private Wojtek.
During the Battle of Monte Cassino in Italy a legend was born. Observing his Polish comrades carrying heavy boxes of artillery shells and other munitions to the front Wojtek pitched in by carrying munitions to the front as well. Many soldiers attested to seeing him in action.
After the war ended his company was assigned to a displaced persons camp in Scotland. His presence acted as a buffer to ease relations between the now homeless Poles and their provincial Scots hosts. Eventually as his comrades scattered he was given a new home in the Edinburgh Zoo - where his former comrades would visit him.
The first two thirds of this book covers the life of Wojtek and the final third of the book covers Orr's quest to create a fitting tribute to the bear soldier. An epilogue by Neal Ascherson discusses Poland's role in World War II and how Polish soldiers ended up in the Middle East. Thanks to brother Steve for sending me the book.
links here here here here
It is a great story. In 1942 the 22nd Company, Polish Army Service Corps (Artillery) while marching through Persia adopted a Syrian brown bear whom they named Wojtek (pronounced Voy-chek). Over the next three years as they traveled through the Mideast to Italy and eventually to Scotland Wojtek traveled with them. (see here for their travels.) Knowing no other life Wojtek adopted astonishingly domestic and nearly human behavior. It was frequently said that he thought of himself as one of the company. He lived in tents with them, drank beer and smoked cigarettes with them, wrestled, and responded to their Polish language. In order to get him rations the Poles registered him as Private Wojtek.
During the Battle of Monte Cassino in Italy a legend was born. Observing his Polish comrades carrying heavy boxes of artillery shells and other munitions to the front Wojtek pitched in by carrying munitions to the front as well. Many soldiers attested to seeing him in action.
After the war ended his company was assigned to a displaced persons camp in Scotland. His presence acted as a buffer to ease relations between the now homeless Poles and their provincial Scots hosts. Eventually as his comrades scattered he was given a new home in the Edinburgh Zoo - where his former comrades would visit him.
The first two thirds of this book covers the life of Wojtek and the final third of the book covers Orr's quest to create a fitting tribute to the bear soldier. An epilogue by Neal Ascherson discusses Poland's role in World War II and how Polish soldiers ended up in the Middle East. Thanks to brother Steve for sending me the book.
links here here here here
Voter Fraud-kenstein
This morning I caught a few minutes of a CSPAN panel featuring Charlie Cook of the Cook Political Report discussing the upcoming election. Specifically I found interesting a comment he made about voter fraud (see here 54:00 - 55:00) During the Bush years the DOJ found less than one case of voter fraud per state per year. That would come out to 400 cases out of approximately 146MM registered voters. Here are some stats on voter fraud from Mother Jones. Obviously not an unbiased source but I have no reason to believe the numbers are incorrect. And here is an article on voter fraud fraud which discusses where some of the allegations are coming from. However type "voter fraud" into Google and we get 74,000 hits. In some circles it is accepted fact that voter fraud is a massive problem. Here is an sample from noted right-wing website FreeRepublic. I think Cook gets it right - voter fraud is for people who cannot admit that their side can lose elections - either their candidate wins or there must have been voter fraud.
That got me thinking; in the US citizens over the age of 18 can (still) vote. In my home city of Chicago it is also a well established fact that the dead can vote (see here for a discussion of the 1960 election). But can the undead vote ? After discussing this matter with brother Alan I think we have come to the conclusion that
That got me thinking; in the US citizens over the age of 18 can (still) vote. In my home city of Chicago it is also a well established fact that the dead can vote (see here for a discussion of the 1960 election). But can the undead vote ? After discussing this matter with brother Alan I think we have come to the conclusion that
- werewolves yes
- vampires yes but only absentee
- zombies yes (they generally vote R due to a lack of brains har har)
- Frankenstein monsters no
Time for someone to climb back under his rock
The National Enquirer view of Politics
Noonan: When American's Saw the Real Obama
Readers of the Visible Foot may recall my frustration with Peggy Noonan (see here) due to her backward method of viewing politicians and public policy. Rather than deciding which politicians are good/bad based upon the goodness/badness of the policies that they espouse - she decides which policies are good/bad based upon the goodness/badness of the person supporting them. How does she decide which politicians are good/bad? She somehow distills this from their personality and style. I recently happened upon this article and it is a perfect example. Notice the lack of any discussion of policies? The focus is completely on the President's style and personality. For comparison here is Noonan's 2008 endorsement of Barack Obama for President - also completely devoid of any discussion of policies.
Noonan's View of the World:
Readers of the Visible Foot may recall my frustration with Peggy Noonan (see here) due to her backward method of viewing politicians and public policy. Rather than deciding which politicians are good/bad based upon the goodness/badness of the policies that they espouse - she decides which policies are good/bad based upon the goodness/badness of the person supporting them. How does she decide which politicians are good/bad? She somehow distills this from their personality and style. I recently happened upon this article and it is a perfect example. Notice the lack of any discussion of policies? The focus is completely on the President's style and personality. For comparison here is Noonan's 2008 endorsement of Barack Obama for President - also completely devoid of any discussion of policies.
Noonan's View of the World:
- Good Personality & Style -> Good Person -> Support Policy
- Bad Personality & Style -> Bad Person -> Do Not Support Policy
- Good Policy -> Support Person
- Bad Policy -> Do Not Support Person
Monday, October 22, 2012
BP to sell its share of TNK-BP to Rosneft
NY Times story
"MOSCOW — BP’s board has approved an offer from the Russian state oil company, Rosneft, to buy most of BP’s business in Russia for cash and shares in Rosneft...For BP, the sale comes as the biggest step yet in Mr. Dudley’s “shrink to grow” strategy...selling older, less profitable fields and concentrating on exploration... The Russian holdings were the largest that fit the category for potential sales, comprising mostly aging oil fields in Siberia with little potential for new output. Still, they account for about 25 percent of BP’s global production, or
about as much oil as BP pumps in the United States, including Alaska...TNK-BP is the third-largest Russian oil company; if Rosneft buys out
both BP and the handful of Russian billionaires who control the other
half of TNK-BP, Rosneft will become the world’s largest publicly traded
oil company, with production of about 4.5 million barrels a day. "
Sunday, October 21, 2012
Credit Default Swaps for Dummies: Part V - Correlation and Synthetic CDOs
In Part I of Credit Default Swaps for Dummies we explained single name CDS.
In Part II of Credit Default Swaps for Dummies we explained CDS indices.
In Part III of Credit Default Swaps for Dummies we explained asset securitization.
In Part IV of Credit Default Swaps for Dummies we explained synthetic CDOs.
In Part V of Credit Default Swaps for Dummies we will demonstrate how correlations between defaults affect the pricing of synthetic CDOs.
In Part II of Credit Default Swaps for Dummies we explained CDS indices.
In Part III of Credit Default Swaps for Dummies we explained asset securitization.
In Part IV of Credit Default Swaps for Dummies we explained synthetic CDOs.
In Part V of Credit Default Swaps for Dummies we will demonstrate how correlations between defaults affect the pricing of synthetic CDOs.
Assume we are asked to insure 100 USD of company A's debt for one year. And assume that we know that the probability that A defaults on her debt over the next year is 50% and if she defaults it will be a full default with 0 USD recovered. Assume that you are willing to insure her for fair value. How much would you require to be paid to write the insurance? There is a 50% chance that A will not default in which case you pay 0 USD. There is also a 50% of A defaulting and you end up having to pay 100 USD. So the premium that you would require to write this insurance would be 50% * 0 USD + 50% * 100 USD = 50 USD. That is just a very simple CDS pricing model.
Now assume that we are asked to insure 100 USD of company B's
debt for one year. Assume we know that
the probability that B defaults on his debt over the next year is 50% and if he
defaults it will be a full default with 0 USD recovered. And assume that you are willing to insure him
for fair value. How much would you
require to be paid to write the insurance? Again you would require to be paid a premium of 50% * 0 USD + 50% * 100 USD = 50 USD.
Now assume you are asked to create an index CDS which is
made up of 100 USD of company A and 100 USD of company B. And assume that you are willing to insure
for fair value. How much would you
require to be paid to write insurance on this basket?
The index CDS is just the sum of the two individual CDSs above. So you would require to be a paid a premium of 50 USD to insure company A's part of the basket and 50 USD to insure company B's part of the basket or 50 USD + 50 USD = 100 USD to insure this basket. So good so far.
Now let’s create a CDO.
Let’s create an underlying pool of 100 USD of A and 100 USD of B. Structure it with two tranches; senior and junior. The first 100 USD of
defaults will go to the junior tranche.
Any defaults above 100 USD will go to the senior tranche. Now how much premium would you require to
agree to write the junior or senior tranche?
It turns out that it is very dependent on what you assume about the
correlations in the defaults. Let's look at a few extreme cases;
Correlation in Defaults = 100%
First assume that there is a 100% correlation between the
defaults of A and B. So if A defaults
then B defaults, and if B defaults then A defaults. Let’s look at the possible outcomes and their
associated probabilities
(1)
50 % chance: A and B both default. Loss to the pool (if it did occur) = 200 USD.
The junior tranche absorbs 100 USD of
losses and the senior tranche absorbs 100 USD of losses.
(2)
50% chance:
neither A nor B defaults. Loss to
the pool = 0 USD. Neither the senior nor
the junior tranche absorbs any losses.
(3)
0% chance:
A defaults and B does not default.
Loss to the pool (if it did occur) 100 USD. The junior tranche absorbs 100 USD in
losses. The senior tranche absorbs 0 USD
in losses.
(4)
0% chance:
B defaults and A does not default.
Loss to the pool (if it did occur) 100 USD. The junior tranche absorbs 100 USD in
losses. The senior tranche absorbs 0 USD
in losses.
So what is the fair value premium to write each of the two
tranches? To calculate the expected loss from writing insurance on one of the tranches calculate ProbabilityOf(case 1)*Payout(if case 1 occurs)+ProbabilityOf(case 2)*Payout(if case 2 occurs)+ProbabilityOf(case 3)*Payout(if case 3 occurs)+ProbabilityOf(case 4)*Payout(if case 4 occurs). So the expected loss from writing
the junior tranche is 50% * 100 USD + 50% * 0 USD + 0% * 100 USD + 0% * 100 USD
= 50 USD. Similarly the expected loss
from writing the senior tranche is 50% * 100 USD + 50% * 0 USD + 0% * 100 USD +
0% * 100 USD = 50 USD. So if the correlation in defaults is 100% then
you would require 50 USD to write either the senior or junior tranches. This may look a bit strange but the first to default structure results in the two tranches senior and junior paying out equally in two situations - if either both A and B default or if neither A nor B default. By our assumption about correlations A defaults if and only if B defaults. Hence senior and junior tranches have the same expected payouts.
Correlation in Defaults = -100%
Now let’s assume that there is a -100% correlation between
the defaults of A and B. So if A
defaults then B does not default, and if B defaults then A does not
default. Let’s look at the possible
outcomes and their associated probabilities
(1)
0 % chance:
A and B both default. Loss to the
pool (if it did occur) = 200 USD. The junior
tranche absorbs 100 USD of losses and the senior tranche absorbs 100 USD of
losses.
(2)
0% chance:
neither A nor B defaults. Loss to
the pool (if it did occur) = 0 USD.
Neither the senior nor the junior tranche absorbs any losses.
(3)
50% chance:
A defaults and B does not default.
Loss to the pool (if it did occur) 100 USD. The junior tranche absorbs 100 USD in
losses. The senior tranche absorbs 0 USD
in losses.
(4)
50% chance:
B defaults and A does not default.
Loss to the pool (if it did occur) 100 USD. The junior tranche absorbs 100 USD in
losses. The senior tranche absorbs 0 USD
in losses.
So what is the fair value premium to write each of the two
tranches? The expected loss from writing the junior
tranche is 0% * 100 USD + 0% * 0 USD + 50% * 100 USD + 50% * 100 USD = 100
USD. Similarly the expected loss from
writing the senior tranche is 0% * 100 USD + 0% * 0 USD + 50% * 0 USD + 50% * 0
USD = 0 USD. So if the correlation in
defaults is -100% then you would require 100 USD to write the junior tranche but
0 USD to write the senior tranche. This
looks a bit strange but in this case the defaults never reach up to the
senior tranche so he never ends up paying out.
Correlation in Defaults = 0%
Now let’s assume that there is a 0% correlation between the
defaults of A and B. So the
probabilities of A and B defaulting are independent- meaning that A defaulting
tells you nothing about whether B defaulted and visa-versa. Let’s look at the possible outcomes and their
associated probabilities
(1)
25 % chance:
A and B both default. Loss to the
pool (if it did occur) = 200 USD. The
junior tranche absorbs 100 USD of losses and the senior tranche absorbs 100USD
of losses.
(2)
25% chance:
neither A nor B defaults. Loss to
the pool (if it did occur) = 0 USD.
Neither the senior nor the junior tranche absorbs any losses.
(3)
25% chance:
A defaults and B does not default.
Loss to the pool (if it did occur) 100 USD. Junior tranche absorbs 100 USD in losses. The senior tranche absorbs 0 USD in losses.
(4)
25% chance:
B defaults and A does not default.
Loss to the pool (if it did occur) 100 USD. Junior tranche absorbs 100 USD in losses. The senior tranche absorbs 0 USD in losses.
So now what is the fair value premium to write each of the
two tranches? The expected loss from
writing the junior tranche is 25% * 100 USD + 25% * 0 USD + 25% * 100 USD + 25%
* 100 USD = 75 USD. Similarly the
expected loss from writing the senior tranche is 25% * 100 USD + 25% * 0 USD +
25% * 0 USD + 25% * 0 USD = 25 USD. So
if the correlation in defaults is 0% then you would require 75 USD to write the
junior tranche and 25 USD to write the senior tranche.
Summary of the above results
Correlation in Defaults
|
Junior Tranche Required Premia
|
Senior Tranche Required Premia
|
100%
|
50
|
50
|
0
|
75
|
25
|
-100%
|
100
|
0
|
In each of the above cases the probability of A
defaulting was 50% and the probability of B defaulting was 50% but the pricing
of the two tranches of the CDO turned out to be very different depending on what we
assumed about the correlations in the defaults.
Obviously this was a very contrived example but it demonstrates why
correlations matter. The problem really
comes in determining the correlations among defaults. Accurate probabilities of default can be
difficult to predict but correlations between events of default are even
more difficult to predict especially when you are dealing with small probabilities to start with.
Now that you are an expert on Credit Default Swaps and synthetic CDOs lets return to the original topic - how JPMorgan lost billions see here here here
For more on these topics see
- Understanding Credit Derivatives and Related Instruments - Antulio Bonfim
- The Credit Default Swap Basis - Moorad Choudhry
- CQF Module 5
- Markit Credit Indices Primer
- Structures Credit Products - Moorad Choudhry
- Credit Default Swaps - Wikipedia
- Collateralized Debt Obligations - Wikipedia
Credit Default Swaps for Dummies: Part IV - Synthetic CDOs
In Part I of Credit Default Swaps for Dummies we explained single name CDS.
In Part II of Credit Default Swaps for Dummies we explained CDS indices.
In Part III of Credit Default Swaps for Dummies we explained asset securitization.
In Part IV of Credit Default Swaps for Dummies we will explain what a synthetic CDO is.
Creating a synthetic corporate bond from a CDS
First recall what a CDS is (see Part I if you forgot how these work). Say you are a bank and you have made a five year loan of 1MM USD to company X. You are ok with lending out 1MM for five years and you are willing to assume the interest rate risk of this loan. But you don't like the credit risk of company X - you think there is a possibility that they may default on their debt. What can you do? One solution is to try to sell off the the debt to someone else - but that may be cost prohibitive. A second solution might be to purchase a CDS for 1MM of five year protection on company X. Each year you would pay a premium (aka interest rate or coupon) to the CDS seller. If company X has a "credit event" within the next five years then you would receive an insurance payment that would make up the difference between the face value of the debt (1MM USD) and the value of the debt after the credit event.
Now let us turn it around and assume that you are a portfolio manager and you have 1MM USD to invest for five years. You could buy US Treasury debt and receive an interest rate of 0.625% - but that is pretty crummy. In order to receive a higher interest rate you are going to need to be willing to assume some credit risk. You like company X's business plan and you think the chance of them defaulting on their debt is pretty miniscule. You look at the bond market and see that company X's corporate bonds are paying an interest rate of 3.625%. If you were to purchase this bond you would effectively receive 0.625% (the Treasury rate) to compensate you for the time value of money and for the interest rate risk, and you would be receive 3% to compensate you for the risk of a possible default by company X.
As a portfolio manager could purchase 1MM USD face of five year bonds from company X ...or..you could synthetically recreate this same cash flows using US Treasuries and CDSs. To do the latter you would purchase 1MM USD face of five year US Treasuries- that will get you an interest rate of 0.625% for the next five years to compensate you for time value of money and for the interest rate risk. Then you would sell a 1MM USD face of five year protection on company X (as a CDS) to the bank. If company X were to default on its debt then you would have to pay the bank the difference between the 1MM USD of face and the price of the debt post default. So in the case of a default you would absorb the default loss just as if you directly held the bonds of company X.
How much in premium should you demand to be paid in order to write this CDS protection? By the principle of no-arbitrage one would assume that the portfolio of Treasury+CDS should pay the same interest rate as company X's bonds (3.625%). If the Treasury+CDS portfolio were to pay less than 3.625% then no one would want to sell the CDS protection to the bank since they could instead just buy the corporate bonds, bear the same default risk and get a higher interest rate. If the Treasury+CDS portfolio paid more than 3.625% then other agents who currently owned company X bonds should be willing to sell their bonds and instead synthetically
recreate the bonds by purchasing Treasuries and selling CDS protection
to the bank. Since company X's corporate bonds pay 3.625% and Treasuries pay 0.625% then you should require 3% coupon to write the CDS protection.
Cash CDOs
Now let us extend the principle to a pool of loans. First we want to quickly review cash securitization (see Part III if you forgot how these work). Say you are bank and you hold five year bonds for 1MM USD face for each of 1000 companies. You are ok with loaning out 1BB USD and you are willing to assume the interest rate risk of making these loans but you don't like bearing the credit risk of these bonds, now what are your alternatives? One solution would be to sell these bonds off individually - but that may be prohibitive as you would have to find takers for each of the 1000 bonds.
A second solution would be to securitize the bonds. You the bank could create an SPV. The SPV would issue its own debt to the public and use the proceeds to purchase the bonds from you the bank. The pool of bonds would pay interest and principal to the SPV and the SPV would distribute these cash flows to the investors in the SPV. There are a number of different ways that the SPV could structure their payouts.
The first structure that we are interested in is a direct pass-through. In this case there would be a single class of SPV debt. The SPV would pass interest and principal cash flows through to investors in proportion to the percent of the SPV's debt that the investor holds. In the case of defaults from the underlying pool an investor would absorb a share of defaults in proportion to the percent of the SPV's debt that he or she holds.
The second structure that we want to focus on is the Collateralized Debt Obligation (CDO). In this case the SPV issues multiple classes of debt ordered in a hierarchy (super-senior, senior, mezzanine, junior, equity, etc..). Each class of debt gets allocated a share of the principal. The
most junior class of debt will absorb all
defaults until the principal of that class is depleted. After the
principal of the most junior class is exhausted then the next most
junior class will absorb all defaults until the principal of that class
is depleted
and so on. The lower is a class in the hierarchy the higher the interest rate
that it will receive but the more default risk it will be exposed to.
The most senior tranches will be exposed to very little default risk but
in return they receive the lowest interest rate. We refer to this as a "cash" CDO because it involves a securitization of the actual cash bonds.
Synthetic CDOs
An alternative way for the bank to insure the pool of bonds is with a synthetic CDO. Recall that earlier in this post we demonstrated how we could synthetically recreate a corporate bond using US Treasuries plus a CDS. Could we synthetically recreate a "cash" CDO using US Treasuries plus some sort of CDS like structure? The answer is yes.
A synthetic CDO would be structured as follows. The bank forms an SPV. The SPV sells CDS insurance to the bank on each of its 1000 names. In return the bank pays an insurance premium (aka coupon or interest rate) to the SPV. The SPV distributes this premium to one or more investors. If one of the bonds goes into default the investors are responsible to pay the SPV who in turn pays the bank the difference between the face value of the debt and the after default value of the debt.
Just as we saw with simple cash securitizations (see Part III) there are a number of ways that the premium payments and insurance obligations can be divided amongst the investors. The most simple structure for a synthetic CDO is as a synthetic pass-through. In this case each each investor would be assigned a share of the total face value of the pool. He would receive a share of the premium payments proportional to his share of the pool. In the case of default he would be responsible to cover a share of the defaulted bonds proportional to his share of the pool. This is structure is fairly similar to an index CDS (see Part II) if the bank were to choose the makeup of the index.
It should be obvious that we can recreate the "cash" pass-through security by combining US Treasuries with a synthetic pass-through. Assume that a five year US Treasury yields 0.625%. And assume that a five year cash pass-through security with an underlying pool of 1MM USD each of 1000 names pays 3.625%. Of this 3.625% the first 0.625% compensates us for the time value of money and interest rate risk. The remaining 3% compensates us for the default risk that we take on this pool. Now assume that the bank were to issue a five year synthetic pass-through with an underlying pool of 1MM each of the same 1000 names. If we were to purchase 1BB USD face of five year US Treasuries and then sell protection via this synthetic pass-through then we would be exposed to the same time value of money, interest rate risk, and default risk as if we had purchased the original cash pass-through. The cash pass-through pays 3.65% and the Treasuries yield 0.625% so the synthetic pass-through must yield 3% else there would exist unexploited arbitrage opportunities between the cash and synthetic markets.
Alternately we could structure the SPV as a credit tranched synthetic CDO. In this case the the SPV would create different classes (super-senior, senior, mezzanine, junior, equity, etc). A class would be defined by its responsibility for paying out insurance in the case of defaults in the underlying pool and by the premium it receives as compensation. Typically the responsibility for paying insurance is defined in terms of attachment and detachment points which specify what is the minimum level of defaults at which that class will start paying and what is the maximum level of defaults that that class will be responsible for.
Since this is a bit complicated I will fully spell out an example. Assume our bank has 1MM of debt from each of 1000 names. 1BB in total for the pool. They create a synthetic CDO with the following structure. An SPV is created which writes protection on the full 1BB in debt. The SPV creates four tranches; equity, junior, mezzanine, and senior. They are defined as follows:
- equity tranche - insures the first 50MM in defaults from the pool.
- junior tranche - insures the next 100MM in defaults from the pool.
- mezzanine tranche - insures the next 200MM of defaults from the pool.
- senior tranche - insures the last 650MM of defaults from the pool.
The below diagram illustrates each tranches responsibility for paying out insurance in the case of defaults in the pool.
The bank pays the SPV a premium of 3% to insure 1BB of debt or (3% * 1BB = ) 30MM per year. This premiums gets divided between the four tranches. We will assume that the tranches receive premium as follows.
- equity tranche - receives 20MM or (20MM / 50MM = ) 40% return per year
- junior tranche - receives 7MM or (7MM / 100MM = ) 7% return per year
- mezzanine tranche - receives 2MM or (2MM / 200MM = ) 1% return per year
- senior tranche - receives 1MM or (1MM / 650MM = ) 0.15% return per year
Below is a diagram of the full transaction
We previously showed that you can mimic a long position in a corporate bond by buying a Treasury and selling CDS insurance on the same name. And we asserted that you can mimic a long position in a cash pass-through by buying a Treasury and selling insurance on the same pool via a synthetic pass-through. It should not come as a surprise that we can mimic a cash CDO by buying a Treasury and selling insurance on the same pool via a properly structured synthetic CDO. In addition we can mimic the return on each tranche of a cash CDO by correctly structuring a synthetic CDO on the same pool.
Finally - although the examples that I have given above assumed that we used the synthetic CDO to fully insure the pool, in many cases a synthetic CDO will be written to only cover a small percentage of defaults (say the first 10 - 20%) in the pool. In this case each tranche will only be responsible for a few percentage points worth of defaults in the pool. The idea there is that in a large diversified pool it is very likely that there will be some defaults but it is very unlikely that a large percent of the pool will default (although that depends on the quality of the debt in the pool).
Tuesday, October 16, 2012
But what would Ayn Rand say on the morality of phony altruism?
Here is Rep. Paul Ryan's ideological inspiration Ayn Rand on the morality of altruism. And then there is this...
Washington Post is reporting "Ryan had stopped by the soup kitchen for about 15 minutes on his way to the airport after his Saturday morning town hall in Youngstown. By the time he arrived, the food had already been served, the patrons had left, and the hall had been cleaned. Upon entering the soup kitchen, Ryan, his wife and three young children greeted and thanked several volunteers, then donned white aprons and offered to clean some dishes. Photographers snapped photos and TV cameras shot footage of Ryan and his family washing pots and pans that did not appear to be dirty."
"Brian J. Antal, the president of an Ohio charity for the homeless, is objecting to a photo op staged by the Romney campaign on Saturday that featured vice presidential running mate Paul Ryan washing pots and pans that had already appeared to be clean, Felicia Sonmez of the Washington Post reports:" here is the photo in question
Actually maybe Rand would have been ok with this because
Washington Post is reporting "Ryan had stopped by the soup kitchen for about 15 minutes on his way to the airport after his Saturday morning town hall in Youngstown. By the time he arrived, the food had already been served, the patrons had left, and the hall had been cleaned. Upon entering the soup kitchen, Ryan, his wife and three young children greeted and thanked several volunteers, then donned white aprons and offered to clean some dishes. Photographers snapped photos and TV cameras shot footage of Ryan and his family washing pots and pans that did not appear to be dirty."
"Brian J. Antal, the president of an Ohio charity for the homeless, is objecting to a photo op staged by the Romney campaign on Saturday that featured vice presidential running mate Paul Ryan washing pots and pans that had already appeared to be clean, Felicia Sonmez of the Washington Post reports:" here is the photo in question
Actually maybe Rand would have been ok with this because
- the intent of Ryan's action was not to help any homeless person
- the effect of Ryan's action was not to help any homeless person
Sunday, October 14, 2012
Nobel Prediction
Tomorrow the Nobel Prize for Economics will be announced. Names being bantered around include
Late Note: hmm my macro-finance-centric oversight. I assumed Shapley had already won a Nobel Prize (maybe at the same time as Aumann won). Well he is certainly deserving of one. I don't really know anything about Roth's work other than he has co-authored a number of papers with my former classmate Tayfun Sonmez on applications of matching problems.
- Jean Tirole - for contributions to a number of areas including industrial organization, auction theory, regulation. Co-winner could be Fudenberg.
- Eugene Fama - efficient market hypothesis the idea that asset prices fully incorporate all available public information
- Robert Shiller - why asset prices are not efficient ie Fama is wrong. Co-winners could be Lo & McKinley or Schleifer & Vishny.
- David Romer - endogenous growth theory old growth theory said that the rate of growth of an economy is driven by the growth rates of population and technology - but assumes growth rates of population and technology are fixed outside of the model. New growth theory says that the growth rate of technology is determined by firms decisions on how much to invest in R&D and new technologies. Co-winners could be either Aghion & Howitt or Grossman & Helpman.
- Robert Barro - for contributions to both rational expectations and new growth theory
- Fischer, Gray, and Taylor for developing rational expectations models with sticky prices and wages.
- Calvo - menu costs. Co-winners could be Taylor Akerlof (already won one for lemons) Yellen and Mankiw.
- Stiglitz already won - so maybe his partner Dixit for monopolistic competition.
- Bernanke Gertler Gilchrist - models in which non-trivial financial sectors accelerate shocks
Late Note: hmm my macro-finance-centric oversight. I assumed Shapley had already won a Nobel Prize (maybe at the same time as Aumann won). Well he is certainly deserving of one. I don't really know anything about Roth's work other than he has co-authored a number of papers with my former classmate Tayfun Sonmez on applications of matching problems.
September's Doctored Non-Farms Payrolls
Rick Santelli ranting about jobs numbers
If I posted a link every time Rick Santelli of CNBCrazy goes off on some nutty rant the Visible Foot would get pretty repetitive. I just thought it was ironic that one day after I post about his bad economic analysis he goes off on this rant.
Santelli was was not the only one though who saw through these obviously "Unbelievable jobs numbers". Interestingly, no evidence is offered as to why the numbers are wrong except that they must be wrong because they seem to favor the President near election time. Here Menzie Chen actually looks at the data to see if there is anything out of the ordinary - and concludes it is well within statistical norms. Of course there was never any serious evidence otherwise.
What's really pathetic about the situation is that FOXers don't even realize that the numbers that they are accusing the BLS of doctoring are not that good even in their "doctored" form. The civilian employment to population ratio is essentially unchanged this year - and about 4 points off its 2007 high. Tim Duy has a nice review of the current state of the labor market.
If I posted a link every time Rick Santelli of CNBCrazy goes off on some nutty rant the Visible Foot would get pretty repetitive. I just thought it was ironic that one day after I post about his bad economic analysis he goes off on this rant.
Santelli was was not the only one though who saw through these obviously "Unbelievable jobs numbers". Interestingly, no evidence is offered as to why the numbers are wrong except that they must be wrong because they seem to favor the President near election time. Here Menzie Chen actually looks at the data to see if there is anything out of the ordinary - and concludes it is well within statistical norms. Of course there was never any serious evidence otherwise.
What's really pathetic about the situation is that FOXers don't even realize that the numbers that they are accusing the BLS of doctoring are not that good even in their "doctored" form. The civilian employment to population ratio is essentially unchanged this year - and about 4 points off its 2007 high. Tim Duy has a nice review of the current state of the labor market.
Credit Default Swaps for Dummies: Part III - Securitization
In Part I of Credit Default Swaps for Dummies we explained what a single name CDS is. In Part II of Credit Default Swaps For Dummies
we explained how a CDS Index works. In this section we will (try) to
explain how securitization works. This is a precursor to explaining
synthetic CDOs and why they are so hard to price.
Say you are a savings and loan bank. You take deposits from customers and you lend these deposits to people who want to buy houses. Over the life of the residential home loan the borrower will pay you back the principal on the loan plus interest. Your business is going great but you find that you have more demand for loans than you have deposits. You could try to borrow funds in another manner - say by issuing certificates of deposit (CDs) but there are limits to what you can do there. You are limited by the equity value of the bank as to how much you can borrow and lend. You could issue loans to borrowers and then try to sell the individual loans off to another bank but that can be very time consuming considering that the purchasing bank is not familiar with the borrower. What you need is a way to package up those home loans and sell them off en-masse.
Enter securitization. You (the bank) form a Special Purpose Vehicle (SPV) which is a entity legally separate from the bank. The SPV issues debt to investors and uses the funds from that sale to purchase a pool of home loans from you (the bank). Over time as the home buyers pay back principal and interest those cash flows go to the SPV and are distributed to the investors (debt holders) in the SPV. Any sort of debt can be securitized in this way. Securitizations of mortgage related debt are typically known as Mortgage Backed Securities (MBS). If the mortgages are residential they are known as RMBS and if they are commercial mortgages then they are known as CMBS. If we are securitizing non-mortgage assets (like credit card debt, college loans, car loans, etc) they are known collectively as Asset Backed Securities (ABS).
The big advantage of this type of structure is if there are many borrowers in the pool of loans and the events of default are uncorrelated between borrowers then by the Law of Large Numbers the realized default rate on the pool should approach the population mean default rate. This reduces the need to monitor each borrower. Also instead of one person holding all of those loans the SPV can sell its debt off in small chunks so that many many investors can each own a small interest in a diversified pool of loans. So for example I would not want to take the risk of making one large loan to one borrower but I would be fine with making a small loan to large pool of borrowers since I can gauge the approximate default rate on the pool.
The downsides of this type of structure is (1) post securitization the originating bank will not have ongoing exposure to the borrowers so they may be less careful about who they loan to in the first place (2) by securitizing the loans the bank is no longer responsible for monitoring (3) it is more difficult to adjust terms of the loan when the loan is owned by a large pool as opposed to if it were owned by a single bank. All three of these problems reared their heads during the recent sub-prime debt crisis.
The structure that we describe above specified that the SPV payed out the principal and interest cash flows to its investors - but we did not specify how they get payed out. The most simple structure is a pass-through security. In this case each investor gets a share of any repayed interest and principal proportional to their investment in the SPV. Cash flows "pass through" the SPV directly and proportionally to investors.
In practice though the SPV can issue multiple classes of debt where each class has a claim on different portions of the cash flows. When the underlying pool is made up of residential mortgages this structure is known as a Collateralized Mortgage Obligation (CMO). When the underlying pool contains non-mortgage debt the structure is known as a Collaterized Debt Obligation (CDO).
Four Popular CMO Structures
The most simple CMO structure is the sequential pay structure. In this structure the SPV creates multiple classes of debt; call them A,B,C,D,...Z. An initial amount of principal is allocated to each class. When borrowers start paying back principal initially 100% of the principal payments (and prepayments) are allocated to the A class. After the principal of the A class is fully payed off then the B class starts receiving all principal payments (and prepayments). After the principal of the B class is fully payed off then all principal payments (and prepayments) go to the C class...This pattern continues until all principal is paid off and the Z class is retired. As we move through time each class receives interest payments proportional to the amount of principal still outstanding for that class.
Why would you want to structure a bond like this? It allows lenders to provide funds to the mortgage market but gives them some control over when those loans will be paid off. Say you are a lender who wants to lend for five years. If you look at the entire universe of 30 year fixed rate mortgages most will be payed off at more than five years but a few will be payed off faster. How do you target getting payed back in five years? If you want to loan fairly short term then you purchase an A class. If you want to lend long term then you purchase a Z class. Because the CMO pools a large number of mortgages the percent of the CMO that gets payed off in the first five years should mimic the population average (so says the Law of Large Numbers).
The sequential pay structure works well so long as interest rates stay fairly constant...however if after the CMO is issued interest rates fall we can run into a problem called prepayment risk. When a lender makes a mortgage loan to a homeowner he expects to receive a fixed interest rate and principal payments for the term of the loan. However if interest rates were to fall substantially then the homeowner could choose to refinance his mortgage at the lower interest rate and pay off the principal of the original mortgage loan. This leaves our lender with cash to loan again but interest rates are now lower. Some lenders would be willing to loan to the mortgage market but want to avoid prepayment risk. To solve the problem of prepayment risk two types of structures were created; IO-POs and PACs.
In the IO-PO structure the SPV issues two different classes of debt IO (Interest Only) and PO (Principal Only). All interest payed to the SPV goes to the investors holding the IO class of debt and all principal payed to the SPV goes to investors holding the PO class of debt. If interest rates fall substantially then homeowners refinance their loans and pay off their principal to the CMO pool early. This turns out be good for the PO class. Why? Because the POs were not getting any interest payments anyways. They want to get their money back as soon as possible so they can loan it again. But it is bad for the IOs - because they bought their class of bonds with the expectation of getting a sequence of interest payments - and since the principal got paid off early the IOs are not going to get all the interest that they expected. This makes the price of IO and PO classes extremely sensitive to the level of interest rates.
A better solution for the prepayment problem is called a Planned Amortization Class (PAC). In the PAC structure the SPV creates two classes of debt; a PAC class and a companion (aka support) class. Each class is allocated an initial principal amount. The PAC is promised a fixed series of cash flows. The companion class absorbs all prepayments (up to some level). Each class receives interest on the portion of her principal still outstanding. In return for distributing the prepayment risk in this manner the PAC receives a lower yield than a straight pass-through would and the support class receives a higher yield than a straight pass-through would.
A fourth type of CMO is referred to as credit tranching. Credit tranching was created to solve the following problem. Some lenders may have cash to lend to the mortgage market but they do not want to be exposed to potential homeowner mortgage defaults. In the credit tranche structure the SPV issues multiple classes of debt with a hierarchical structure (super-senior, senior, mezzanine, junior, equity, etc..). Each class of debt gets allocated a share of principal. The most junior class of debt will absorb all defaults until the principal of that class is depleted. After the principal of the most junior class is exhausted then the next most junior class will absorb all defaults until the principal of that class is depleted and so on. The lower is a class in the hierarchy the higher the yield that it will receive but the more default risk it will be exposed to. The most senior tranches will be exposed to very little default risk but in return they receive the lowest yield.
CDOs have analogous structures but the underlying pools are non-mortgage debt (credit cards, auto loans, corporate debt, school loans, etc..). One type of CDO of particular interest is created by credit tranching a pool of corporate debt. We will look at this case more in depth in the next section.
If my explanations were still a bit unclear, this presentation gives some examples. Here is another presentation of similar material.
Say you are a savings and loan bank. You take deposits from customers and you lend these deposits to people who want to buy houses. Over the life of the residential home loan the borrower will pay you back the principal on the loan plus interest. Your business is going great but you find that you have more demand for loans than you have deposits. You could try to borrow funds in another manner - say by issuing certificates of deposit (CDs) but there are limits to what you can do there. You are limited by the equity value of the bank as to how much you can borrow and lend. You could issue loans to borrowers and then try to sell the individual loans off to another bank but that can be very time consuming considering that the purchasing bank is not familiar with the borrower. What you need is a way to package up those home loans and sell them off en-masse.
Enter securitization. You (the bank) form a Special Purpose Vehicle (SPV) which is a entity legally separate from the bank. The SPV issues debt to investors and uses the funds from that sale to purchase a pool of home loans from you (the bank). Over time as the home buyers pay back principal and interest those cash flows go to the SPV and are distributed to the investors (debt holders) in the SPV. Any sort of debt can be securitized in this way. Securitizations of mortgage related debt are typically known as Mortgage Backed Securities (MBS). If the mortgages are residential they are known as RMBS and if they are commercial mortgages then they are known as CMBS. If we are securitizing non-mortgage assets (like credit card debt, college loans, car loans, etc) they are known collectively as Asset Backed Securities (ABS).
The big advantage of this type of structure is if there are many borrowers in the pool of loans and the events of default are uncorrelated between borrowers then by the Law of Large Numbers the realized default rate on the pool should approach the population mean default rate. This reduces the need to monitor each borrower. Also instead of one person holding all of those loans the SPV can sell its debt off in small chunks so that many many investors can each own a small interest in a diversified pool of loans. So for example I would not want to take the risk of making one large loan to one borrower but I would be fine with making a small loan to large pool of borrowers since I can gauge the approximate default rate on the pool.
The downsides of this type of structure is (1) post securitization the originating bank will not have ongoing exposure to the borrowers so they may be less careful about who they loan to in the first place (2) by securitizing the loans the bank is no longer responsible for monitoring (3) it is more difficult to adjust terms of the loan when the loan is owned by a large pool as opposed to if it were owned by a single bank. All three of these problems reared their heads during the recent sub-prime debt crisis.
The structure that we describe above specified that the SPV payed out the principal and interest cash flows to its investors - but we did not specify how they get payed out. The most simple structure is a pass-through security. In this case each investor gets a share of any repayed interest and principal proportional to their investment in the SPV. Cash flows "pass through" the SPV directly and proportionally to investors.
In practice though the SPV can issue multiple classes of debt where each class has a claim on different portions of the cash flows. When the underlying pool is made up of residential mortgages this structure is known as a Collateralized Mortgage Obligation (CMO). When the underlying pool contains non-mortgage debt the structure is known as a Collaterized Debt Obligation (CDO).
Four Popular CMO Structures
The most simple CMO structure is the sequential pay structure. In this structure the SPV creates multiple classes of debt; call them A,B,C,D,...Z. An initial amount of principal is allocated to each class. When borrowers start paying back principal initially 100% of the principal payments (and prepayments) are allocated to the A class. After the principal of the A class is fully payed off then the B class starts receiving all principal payments (and prepayments). After the principal of the B class is fully payed off then all principal payments (and prepayments) go to the C class...This pattern continues until all principal is paid off and the Z class is retired. As we move through time each class receives interest payments proportional to the amount of principal still outstanding for that class.
Why would you want to structure a bond like this? It allows lenders to provide funds to the mortgage market but gives them some control over when those loans will be paid off. Say you are a lender who wants to lend for five years. If you look at the entire universe of 30 year fixed rate mortgages most will be payed off at more than five years but a few will be payed off faster. How do you target getting payed back in five years? If you want to loan fairly short term then you purchase an A class. If you want to lend long term then you purchase a Z class. Because the CMO pools a large number of mortgages the percent of the CMO that gets payed off in the first five years should mimic the population average (so says the Law of Large Numbers).
The sequential pay structure works well so long as interest rates stay fairly constant...however if after the CMO is issued interest rates fall we can run into a problem called prepayment risk. When a lender makes a mortgage loan to a homeowner he expects to receive a fixed interest rate and principal payments for the term of the loan. However if interest rates were to fall substantially then the homeowner could choose to refinance his mortgage at the lower interest rate and pay off the principal of the original mortgage loan. This leaves our lender with cash to loan again but interest rates are now lower. Some lenders would be willing to loan to the mortgage market but want to avoid prepayment risk. To solve the problem of prepayment risk two types of structures were created; IO-POs and PACs.
In the IO-PO structure the SPV issues two different classes of debt IO (Interest Only) and PO (Principal Only). All interest payed to the SPV goes to the investors holding the IO class of debt and all principal payed to the SPV goes to investors holding the PO class of debt. If interest rates fall substantially then homeowners refinance their loans and pay off their principal to the CMO pool early. This turns out be good for the PO class. Why? Because the POs were not getting any interest payments anyways. They want to get their money back as soon as possible so they can loan it again. But it is bad for the IOs - because they bought their class of bonds with the expectation of getting a sequence of interest payments - and since the principal got paid off early the IOs are not going to get all the interest that they expected. This makes the price of IO and PO classes extremely sensitive to the level of interest rates.
A better solution for the prepayment problem is called a Planned Amortization Class (PAC). In the PAC structure the SPV creates two classes of debt; a PAC class and a companion (aka support) class. Each class is allocated an initial principal amount. The PAC is promised a fixed series of cash flows. The companion class absorbs all prepayments (up to some level). Each class receives interest on the portion of her principal still outstanding. In return for distributing the prepayment risk in this manner the PAC receives a lower yield than a straight pass-through would and the support class receives a higher yield than a straight pass-through would.
A fourth type of CMO is referred to as credit tranching. Credit tranching was created to solve the following problem. Some lenders may have cash to lend to the mortgage market but they do not want to be exposed to potential homeowner mortgage defaults. In the credit tranche structure the SPV issues multiple classes of debt with a hierarchical structure (super-senior, senior, mezzanine, junior, equity, etc..). Each class of debt gets allocated a share of principal. The most junior class of debt will absorb all defaults until the principal of that class is depleted. After the principal of the most junior class is exhausted then the next most junior class will absorb all defaults until the principal of that class is depleted and so on. The lower is a class in the hierarchy the higher the yield that it will receive but the more default risk it will be exposed to. The most senior tranches will be exposed to very little default risk but in return they receive the lowest yield.
CDOs have analogous structures but the underlying pools are non-mortgage debt (credit cards, auto loans, corporate debt, school loans, etc..). One type of CDO of particular interest is created by credit tranching a pool of corporate debt. We will look at this case more in depth in the next section.
If my explanations were still a bit unclear, this presentation gives some examples. Here is another presentation of similar material.
Subscribe to:
Posts (Atom)