My photo
My investing philosophy mostly centers around the Value discipline and GARP- Growth at a Reasonable Price. This blog includes commentary on market conditions as well as fundamental analysis of specific companies. Graduated from Rhodes College with a degree in Business with concentration in Finance & Marketing. Currently working on obtaining the CFA designation. Previously worked in Mortgage Trading for a major bank. Use MS Excel extensively for developing investment models, notably valuation models based on DCF methods.

Friday, August 10, 2007

Sub-Prime Fears Affect Prime Borrowers

Sub-prime woes are showing signs of pinching the most credit-worthy borrowers as jumbo-conforming spreads on prime mortgages have sky rocketed. Mortgages are considered “Jumbo” for amounts over $417,000. Fannie Mae and Freddie Mac will only securitize mortgages with conforming loan balances into an Agency MBS which trade in a highly liquid market. The spread has generally been about 1/8 – 1/4 pt higher in rate for those High balance loans. As of this week, the spread has risen to roughly 80 bps (0.8%) according to That’s a national average, for some lenders it’s much worse. The WSJ reported last week that a broker informed them that Wells Fargo had raised their jumbo 30yr rate from 6.875% to 8% effectively suspending production for that loan program. Why? Ostensibly from an absence of demand on Wall Street.

Banks seek to match the duration or “Life” of assets and liabilities. It’s very risky to borrow short-term, i.e. deposits, CDs, etc. and lend long-term for 15-30 years. Generally, financial institutions have a mortgage company/division responsible for originating mortgages and then selling to the secondary market. The mortgage division sells short maturity adjustable-rate mortgages to the bank to satisfy their loan investment portfolio objectives. Longer maturity conforming FRM are assembled into pools that FNMA, FHLC, or GNMA securitizes into MBS which mortgage bankers then sell to Wall Street. Non-conforming mortgages: jumbo, Alt-A, zero-down etc. are bulked up and then sold to Wall Street firms who use that paper to securitize “private-label MBS” or create CMO/CDO structures they offer to investors such as hedge funds.

Agency MBS is an extremely liquid market. Non-conforming mortgages are much less liquid since they are not securitized by a quasi-government agency so finding buyers can be difficult. I worked on a mortgage-trading desk for one of the top 15 largest banks for three years beginning back in 2004. Finding buyers for non-conforming paper was not hard then. Actually, Wall Street firms, especially those in the currently in the headlines, would be beating down our door for as many mortgages they could get their hands on. Lately it appears phones on trading desks at mortgage companies are silent. Certainly the case for Wells Fargo since it’s essentially curtailing jumbo mortgage production. Lenders do not want to originate mortgages they will not be able to sell.

Since most prime jumbo mortgages are underwritten according to Fannie Mae guidelines, the credit risk between conforming and jumbo loans are the same. That’s why borrowers could usually get almost same rate on a $416k loan as they could a $418k loan. Any spread reflects the liquidity premium of the loan, not the default risk. Now, we have seen that liquidity premium spike affecting the most creditworthy borrowers.

The credit risk of conforming MBS and jumbo MBS is significantly different. Agency MBS essentially has no credit risk since they are assumed to be backed by the Federal Government. Jumbo MBS or “Private Label” are backed by whichever Wall Street firm (and mortgage insurers) securitizes them. Thus, the credit risk associated stems from the issuer of the jumbo paper not the underlying loans per se. The credit risk on the mortgages are equal, yet the MBS credit risk is quite different.

In the past, investors assumed that jumbo MBS had minimal credit risk because the financial institutions backing them had enough capital to cover any defaults and that they were “too big to fail.”
Now, given the sub-prime fears, investors worry that non-prime defaults will hamper the backing institution’s ability to cover defaults on prime mortgages. Investors are calling into question the “too big to fail” cliché. Wall Street firms will find it difficult to sell additional mortgage-backed products to investors who are already worried about the fate of the current mortgage fund holdings.

With respect to jumbo spreads, it can be said that the sub-prime mess is spilling into the prime mortgage market. Homeowners trying to sell above $417k face significant challenges opposed to selling at a lower price. Sellers who have been asking slightly higher than the conforming limit may be forced to lower their price causing further downward pressure on home values. Think about how many homes were purchased above $417 during the explosion in home prices. Think about how many of those inflated home values were beyond the actual reach of buyers, but were purchased with a non-traditional mortgage. Interest-only, option payment, and adjustable-rate mortgages with low teasers allowed borrowers to afford expensive homes due to the low payments required during the loan’s early stages. As rates reset higher and monthly payments increase, borrowers will be more likely to default. To avoid foreclosure borrowers will try to sell at the amount owed on their mortgage, but with falling home values and ubiquitous home supply, it will be challenging for sellers to get what they owe.


monk said...

I'm new to blogs and the MBS crisis, so please be patient. I came across your blog when I Goggled "MBS discount rates". I'm trying to better understand the MBS crisis and thought it best to go right to the heart of the matter - pricing of the sold loan. For the sake of this post, assume "loan" and "MBS" are the same. My research seems to indicate that loan purchasers often used the interest rate on the underlying loan(s) less any loan guarantee and servicing fees as their discount rate when evaluating and making a purchase. Using this approach, market value is simply the PV of the loans future net cash flows. What has me puzzled is why a purchaser would use the interest rate on the loan as their discount rate for purchase? The implication is that the loan's interest rate accurately captures the loan's (asset's) risk characteristics. Albeit limited, my experience with mortgage interest rates has been that they're better indicators of current market trends than harbingers of the future. In other words, they’re better marketing tools than accurate indicators of an asset's underlying risk. Your thoughts?

Turley M Muller said...

First, if a buyer valued a loan using the interest rate as the discount rate, then in theory, the price would always equal 100, or par.

I worked on a mortgage trading desk for 3 years and we had to mark our loans to market.
Loans that are more risky carry a higher interest rate because Fannie Mae charges a higher guaranty fee. Standard is 15-20 bps and for higher risk loans FNMA may charge 120bps to insure the loans. Once those loans are secured, there is no default risk to the buyer. Thus, he doesn't receive a higher rate, FNMA does.
The major risk is prepayment risk in agency MBS. Prepayment is an option to the borrower that has value. Its a negative for the investor, since when rates fall borrowers refinance and pay off the higher rate mortgage, thus investors are have to reinvest with a lower rate. Treasury Bond investors don't face that risk, if rates fall, their higher rate bond increases in price and can't be paid off until maturity so the investor enjoys above market interest payments. The discount rate used to value a mortgage is usually the 10yr plus an "option adjusted spread" or another way is the spread to average life which is the expected life of the loan and that is used to pick the treasury rate on the yield curve. Hence, and expected life of 5 years would be compared to a 5yr Treasury Note.
In practice, we used complex computer models to value MBS and loans. Binomial lattices have to be generated to model expected cash flows based on a prepayment model. Valuing mortgages requires a model that captures the optionality. So the discount rate is risk free yield curve plus a spread to account for the option.
Loans with default risk have a higher discount rate to adjust for that risk.
In my experience, buyers weren't using the same discount rate as the interest rate on the asset.

Now, if the firm is holding the asset to maturity (and not held for sale) then they record the asset at face value and offset with the amount of discount/premium which is amortized over the its life. I believe they use the interest rate as the discount rate. Loans held for maturity don't have to be marked to market since there is no intention of selling them. They are supposed to be written down if they have been impaired, but firms only do that if its obvious to every one that there is impairment.

monk said...

Thanks for your thoughts. Recent events seem to indicate that at the core of the MBS crisis lays either a fundamental misunderstanding of the risks associated with this product and how to price those risks, epic level incompetence, wholesale fraud – or some combination thereof.

I continue to focus on discount rates for both the sale and the purchase of the product. Your insight helped me better understand how agency MBS product is priced, but it’s still not clear to me how unsecured product was priced (non-agency MBS). Isn’t this the product referred to as “marketed to model” vs. agency secured product that is “marketed to market”?

Much has been said concerning how modeling errors contributed to the rapid plummet in the MBS market. My experience has been that the most important (and least understood) variable used in both the seller and purchaser’s model is the discount rate used to value the asset(s).

Having an established market helps define current benchmarks (rates), but again, much of the MBS crisis seems to involve the unsecured portion of the MBS market, that which was “marked to model”. How were discount rates derived for sales and purchases in this segment? Also, what’s the total MBS market; what portion is secured and unsecured?

Turley M Muller said...

Thanks for your comments....Yes, the Agency MBS market is basically unaffected, its the "Private Label" or non-agency MBS that is the major concern. Agency MBS have standard underwriting guidelines where as the non-agency mortgages are basically underwritten to the guidelines chosen by the investor.

For example, Bear Stearns, Wells Fargo, Lehman, or Merrill etc. creates relationships with banks/ mortgage lenders for them to originate loans they want to buy. Investors will provide the loan guidelines and the pricing they will pay for a closed loan. The bank/lender then pretty much offers the loan to borrowers at the same price as the investors will purchase (less small servicing release premium). The loan is originated, closed. and then delivered to the investor. The Investors (Wall St) create CMO structures and other complex investment pools they offer to their clients.

The pricing investors offer is basically just a spread to current rates on prime mortgages according to their perceived risk. If a 30yr FNMA mortgage has a 5.5% rate, zero down loans may be 6.25% (+.75%) , Alt-A programs - 7% (+1.5%) and so on…..

Investors screwed up by not pricing in enough risk premium to that rates they were demanding on those loans. Additionally, they should have required tighter credit standards and underwriting. Buyers took on way too much risk for too little reward.

I see what you are saying with regards to discount rates = the Interest rate on the loan, the discount rate may stay the same, but models estimate defaults etc. and reduce future cash flow by the estimated losses. A 10m of mortgages @ 7% would pay 700k in interest a year. If future cash flows are not adjusted for default probabilities, then actual returns will be much lower. If losses are expected to be 200k, then actual cash flow is 500k or 5% interest rate. If you discount the 5% expected actual yield using the original interest rate then the value of the mortgage pool falls way below face value. Adjusting the coupon rate down to a expected yield works just the same as not adjusting the coupon rate (cash flows) and increasing the discount rate to compensate risk. So, using a complex model to forecast future cash flows adjusted for probability of defaults will result in a lower yield that stated on the mortgage. Thus, if investors want to make 7% then they should demand 9% at par value.

Since these mortgages rarely trade, they are marked to model, and the input assumptions can vary considerably. Model assumptions are supposed to be reasonable and reflect the best estimate of value that would be received if they were to sell the assets in the market. Also, valuations are supposed to incorporate recent transaction prices of comparable assets. As defaults mounted, buyers became very fearful, and maybe bid 80 for a face value MBS (at par or 100).

Wall Street firms know if they tried to sell these mortgages in the market the price would take a big hit. So, marking to model allows them to value the mortgages at inflated prices. But, eventually, it becomes completely obvious that the model values are way above probable market value thus forcing them to take a huge write-down on the value.

There is not a liquid market for these non-agency loans. Financial firms originate or correspond the mortgage origination to acquire loans. Then they create real complex structures or “tranches” that they offer to their clients. They plan to hold the assets for the interest income. The fact that there is little trading hides the real value of the assets. Additionally, the mortgage funds are very complex and consist of all types of loans that no investor has a real clear idea what they are investing in and the involved risks.

The agency market is significantly larger than non-conforming mortgages, I am not sure on the exact figures, but maybe 75-25 ??? I would have to look at the total dollar amount of loans originated versus amount securitized by government agency.

monk said...

Thanks again for your help. I borrowed the following from a recent WSJ article (“How Wall Street Stoked the Mortgage Mess”, 6/27/07) to further the consideration of discount rates.

I. Investors
A. Purchase mortgage backed securities (MBS) from investment banks
B. MBS purchased based on investor’s risk/return preference (*Discount rate)

II. Investment Bank (Bear Stearns, Wells Fargo, Lehman, Merrill etc.)
A. Provide to lenders:
1. “Warehouse” line of credit (used to fund mortgages)
2. Mortgage guidelines (documentation, LTV, etc.), pricing (fees) & rates
B. Purchase from lenders
1. Mortgages originated from “warehouse” LOC
C. Securitizes loans by packaging them into mortgage backed bonds (MBS)
D. Sells the MBS to investors based on their risk/return preference (*Discount rate)

III. Brokers & Lenders
A. Conduits for investment banks
1. Initiate mortgage sale
2. Process required paper work
IV. Borrower
A. Receives financing for purchase or re-financing

How did investment banks quantify the perceived risks associated with non-agency MBS product to determine appropriate rate spreads for their loans made to borrowers? These esoteric hybrid products had little if any historical data to determine traditional risk measures (spreads).

Although specificity has been limited, it seems that investors paid the same for agency and non-agency MBS product. Applying a discount rate that recognized the credit rating fraud being perpetrated by the investment banks (credit rating = risk premium) should have resulted in non-agency product purchased at a discount.

Investors should have used a “Marking to model” risk premium! How did investors square their model assumptions with those used by investment banks? I can’t imagine they both recognized the same events and probabilities. Imagine if investors’ purchases were predicated on both complex modeling and realistic discount rates?

Turley M Muller said...

First, I think I may need to clarify/ rectify some of my previous comments that I may have poorly explained and provided meanings not quite in line with my thinking.
Let me back up and clarify something, I may have confused you and anyone else who has followed the thread.

Ok. The mortgage interest rate offered to the borrower is the same as the discount rate which is the investor’s required return. If I am offering a par rate (loan amount = face value of mortgage note, ie no discount points) I have to demand an interest rate equal to my desired return.

Let’s just examine A-rated MBS. The FNMA 5% MBS is trading at par; I buy 10m face value for 10m market price. Couple months later, Inflation has exploded ands have shot up. The FNMA 7% is now trading at par, and I need to sell my 5%. I use 7% as the discount rate to value the 5% cash flows. The selling price is much lower than face value, the price I originally paid. I am forced to sell at a discount and take a loss.

Now, let’s pretend I am packaging up non-agency MBS backed by Alt-A paper. Investors, credit analysts, insurers all believe that the Alt-A should be spread 1% to the A-Prime rate (5%) I offer 6% to the borrower on a Alt-A loan. I package up the loans into a MBS structure that investors are willing to take a 5.75% return on. ( I am taking a cut for my troubles). Well, since these Alt-A loans had new features, historical data is not available to price the risk. After a couple of years, defaults start mounting quick. It’s very apparent that the collateral is much more risky that originally thought, and investors now require 3.5% spread. For new loans, loan officers are now told to charge 3.5% add-on to A-Prime rate of 5% - 8.5% in total. If now the par rate is 8.5%, my 6% loans offered at par, are not worth par. To provide a 8.5% required return I must cut the price on my 6% loans.

Now, to respond to your comments:

You asked:
How did investment banks quantify the perceived risks associated with non-agency MBS product to determine appropriate rate spreads for their loans made to borrowers? These esoteric hybrid products had little if any historical data to determine traditional risk measures (spreads).

I totally agree with you on that, there hardly any historical data to provide estimates of risk since many of these mortgage programs had new, risky features. I think they Wall Street first ascertained what kind of credit rating they agencies would assign on the mortgage collateral. Then they just increased the rate by the risk premium required by the rating.

My opinion:
The credit ratings were way off the mark.

Although specificity has been limited, it seems that investors paid the same for agency and non-agency MBS product. Applying a discount rate that recognized the credit rating fraud being perpetrated by the investment banks (credit rating = risk premium) should have resulted in non-agency product purchased at a discount.

Loans were originated with a risk premium in the rate to compensate for the risk and non-agency guarantee, That’s why Wall Street went on a binge for non-agencies, because of the higher yield. Additionally, Wall Street banks understated the risk to investors, essentially the USP of “higher yield for minimal marginal risk”

My opinion:
The credit ratings were way off the mark. My understanding is that credit agencies provided consulting for the creation and design of these complex collateral structures. After asset structures were build, those same agencies evaluated the credit risk and assigned a rating.
That’s a conflict of interest; if someone is paying an agency for consulting on securitization, the agency better give a good rating in the end if they want to see any future business. Mirrors the Enron case, firms receiving compensation in accounting and reporting, and then auditing the end result. That’s a blatant conflict, and that’s why they changed the law requiring separation of consulting and audit in the accounting process. You remember back in grade school when you took a quiz and then the teacher instructed you to trade papers with your neighbor? Since childhood we have all known the conflict of interest in grading / evaluating one’s own work.

Investors should have used a “Marking to model” risk premium! How did investors square their model assumptions with those used by investment banks? I can’t imagine they both recognized the same events and probabilities. Imagine if investors’ purchases were predicated on both complex modeling and realistic discount rates?

Investor’s for the most part didn’t perform their own valuation methods. They were told by the Investment banks what the credit rating was and interest rate offered. Investors could use the credit rating to compare to other assets of similar trading in the market to obtain a price. Some of these investment banks had in-house hedge funds that were invested in these securities. In that case, investors rely on the managers to make the decisions since that’s why they pay the big money. Ultimately, significant number of investors owned this garbage and were completely unaware of it.

Fund managers probably were much less risk averse than a prudent investor, since it’s their clients’ capital at stake, not necessarily there own. There is an agency issue there.

I think ultimately, the mistake was made on under-estimating defaults. The models are flawed; in order to have meaningful results, considerable sums of historical data is required. Models derive probability estimates from running infinite number of correlation and regression type calculations to determine the factors (or input assumptions) that drive changes in the data.

I witnessed gradual increase in risk-taking for three years while working on the trading desk at Regions.
Since defaults were low, standards were loosened. Defaults continued to remain low, so credit standards became even more aggressive, and still little change in defaults. The lenders/ Wall Street got the false impression that there wasn’t much additional risk in these non-agency type loan programs.

Actually, some borrowers weren’t able service their mortgage debt but they were able ti escape default and foreclosure. Robust appreciation of home values allowed borrowers over-extended to sell their property and pay off the debt before they get behind. Most made money, there more qualified borrowers available since lenders were reducing credit requirements and loan pricing. Home owners lacking the abilty to carry their mortgage were saved by people even more unqualified.

Relaxing credit standards / pricing fueled soaring home values, which mitigated loan defaults, which then urged further loosening in underwriting policy. It’s a self- perpetuating cycle.

Eventually we hit the tipping point, and as home values corrected from exaggerated values, borrowers had less incentive to keep up on their mortgage. Borrowers are not compelled to go the extra mile when their house is worth much less than they owe. With negative equity, the selling price on the home falls short of satisfying the debt obligation and foreclosure ensues.

From what I witnessed, the industry abandoned the focus on credit risk. If the borrower could put 10-30 down, and the appraisal was good, that was all that mattered. Underwriting was based on the collateral, not the borrower. With a down-payment and robust home appreciation, many felt there was no credit risk. If the borrower can’t pay, then he/she can sell the home, or we can foreclose and sell it ourselves possibly making money on an asset we bought at 80 and are selling at 110.
There was an Alt-A program where the borrower was not required to provide documentation, and the rate was determined by credit score. Yet, if the borrower didn’t want to submit a credit score. The could pay a higher rate.

I don’t necessarily disagree underwriting based on collateral as opposed to borrower credit worthiness. I would make a loan to anyone regardless how high the credit risk if I am completely certain that the collateral is good enough to prevent taking a loss.

The error was made in assuming that the collateral would continue to rise (or hold) in value. When the underlying value of the collateral declines, the credit qualities of the borrower are of immense importance.

monk said...

Thanks again, your insights have been very helpful. As more layers are uncovered, the more unnerved I become at what’s transpired. This discussion on discount rates has demonstrated that the market wasn’t driving the evolution of the hybrid mortgage product (i.e. non-agency product), Investment Banks were. The lynch pin is the use and manipulation of Credit Rating Agencies and their ratings.

You and the WSJ have done an excellent job describing the incestuous relationship between Investment Banks’ underwriters and the Credit Rating Agencies. Although the conflict of interest is undeniable, and will undoubtedly receive due attention, the real significance is what the credit ratings represented to the explosive growth of the non-agency MBS market.

The Credit Rating Agencies ratings represented the investor’s de facto discount rate. By insisting that non-agency MBS had similar risk profiles to secured MBS, and therefore deserved similar credit ratings, Investment Bank underwriters and Credit Rating Agencies effectively low balled the Investor’s discount rate.

As you describe, “Investor’s for the most part didn’t perform their own valuation methods. They were told by the Investment banks what the credit rating was and interest rate offered.” I think it’s safe to assume that if the Investor didn’t construct a purchase model, they didn’t construct a discount rate to evaluate the securities' projected flows – they focused instead on the securities’ rate and credit rating.

Again, if Investors had constructed their own discount rates and used them when making investments, their risk profile evaluation of non-agency securities would have resulted in their insistence on a higher risk premium and use of a higher discount rate, that resulted in the purchase of the non-agency securities at a discount to face value.

Simply assigning a “1% spread” to the un-secured product didn’t recognize the glaring differences between the two types of securities. From the very outset, the nu vogue lending standards inherent in the un-secured mortgage products warranted red flags and corresponding higher risk premiums – the lack of historical data notwithstanding.

The Investment Banks’ and Credit Rating Agencies’ quantum leap that a restructuring of credit risk had occurred (i.e. decline in Borrower defaults) due to “robust home appreciation” was simply unfounded (insufficient data to support a wholesale shift in assessing credit risk).

What’s not clear yet is the role of timing. This was a “house of cards” destined to a nasty unraveling. What event(s) caused it to implode? Or as you said – “Eventually we hit the tipping point…” – what was that "tipping point"? Also, what’s next for those still holding bad securities? What affect will it have on the economy?

Turley M Muller said...

Thanks for contributing your meaningful analysis, I think we have raised some very pertinent issues.

I think many of the end investors were unaware of the exact collateral underlying the securities they purchased. If they had known the magnitude of risk that existed, I am sure most would have employed a higher discount rate thus demanding a discount to face value.

I think the tipping point has been the softening of home values. The reason that defaults were very low during the beginning was that borrowers unable to make payment were able to sell their home and pay of the mortgage.
When home prices started to decline, borrowers could only sell their home for less than they owe.
Additionally, some mortgage types had low initial payments, then reset to a higher rate. I think now we are starting to see more defaults due to payments rising.
And we will surely see a lot more.
For the investors who own these securities, they will receive proceeds from the foreclosure and sale of the underlying properties. This will surely add even more pressure to home values, thus the investors will recoup less than they originally invested.

Contact Financial Alchemist

Email: gmail-turley.muller
Phone: (901) 213-8875