Archive for the ‘HECM Prepayments’ Category

FHA’s Underwater Problem – Is the Worst Over?

Monday, January 30th, 2012

FHA recently released another updated Home Equity Conversion Mortgage (“HECM”) loan level data file, this time showing all FHA-insured reverse mortgages originated through November 2011. Once again, prepayment rates declined to new lows: the annual prepayment rate for seasoned HECMs is about 4.7%, compared to the historical average of 6.8%. We have adjusted our HECM “Prepayment By Borrower Age” table accordingly.

Beginning with the prior release (January 2011), FHA added new fields to their data set. This new data enables us to get a sharper picture of the economic state of the HECM program. These include data showing the amount borrowed for each loan, and the year it was drawn. As we explained last April, a good approximation of each loan balance can be calculated by rolling forward the draws, plus estimated accrued interest, MIP, and servicing fees.

Using this same methodology, and applying the data from the most recent FHA dataset, we rolled forward the approximately 580,000 HECM loans currently outstanding. We estimate the total outstanding balance of all HECM loans to be approximately $87.6 billion as of November 2011. This estimate is consistent with FHA’s most recent FY 2011 report.

The dataset also contains the estimated property value at origination for each loan, as well as the mortgaged property’s location: Metropolitan Statistical Area (MSA), State, Zip Code, etc. Once again, for about 92% of the loans in the file, we were able to link these data fields to publicly available data showing historical price levels by MSA, and thereby estimate each loan’s current underlying property value. For the remaining 8%, we used state level home price data. Comparing these property values to the rolled balances, we can estimate the current state of the HECM program with respect to crossover losses.

Our analysis shows the following: approximately 108,000 HECM loans (or 19% of outstanding) are underwater by a total of $4.3 billion, an increase of nearly $1 billion over 10 months. However, those numbers simply compare loan balance to property value; a more accurate measure haircuts the property value to reflect the cost of property disposition. As a practical matter, only the net property value (home price minus the property disposition cost), is available to pay off the HECM loan. Unfortunately, property disposition costs grow significantly as the mortgage balance approaches the crossover point; a haircut of 10% or even 15% is not unreasonable. If the loan is in default or foreclosure, the cost can be much higher. Applying a 10% haircut to property values raises the number of HECMs effectively “crossed over” to 160,000 loans (28% of outstanding), underwater by $6.4 billion, and a 15% haircut to 196,000 loans (34% of outstanding) underwater by $7.9 billion.

Not surprisingly, HECM loans originated from 2005 through 2008 comprise substantially all of these underwater loans. In the 10% haircut scenario, they account for about 88% of the underwater loans. The 2006 and 2007 vintages alone account for nearly 60% of the problem loans.

Because of the 2005-2008 vintages, these numbers will get worse before they get better. The underwater numbers are simply a snapshot as of November 2011, in other words, how much FHA could lose if all the loans paid off at that cutoff date. But these loans are paying off slowly, and will take several years to pay off completely. During those years, accreting loan balances and borrower advances, especially if combined with slow prepayments and a struggling housing market, will make the problem worse. This fact is reflected in FHA’s estimate of a Loan Loss Liability of $10.013 billion (for HECMs originated through FY 2011) in their most recent Annual Management Report. (FHA made upward revisions to their loss estimates for HECMs originated through FY 2008, from their $8.7 billion estimate at the end of FY 2010, $4.8 billion estimate at the end of FY 2009 and $1.5 billion the year before.) We also estimate, based on the FHA dataset, that FHA has already experienced anywhere from $700 million to $1 billion in realized losses from underwater HECMs that have paid off.

Last April we put the HECM program’s net economic value since inception at -$7.3 billion. With declining home prices, (down about 3% in the 12 months ended November 2011), slow prepayments, and accreting loan balances, one might expect FHA’s HECM bottom line to get much worse. But with several months of new loan production of improved HECM products, the overall outlook for FHA’s HECM insurance portfolio avoided further deterioration. With each new loan, FHA is collecting MIP at a higher rate on loans with lower Loan-to-Value ratios. As we noted in our previous blog, FHA reduced its estimate of future losses on its HECM portfolio. Combining FHA’s estimates of what will happen with our estimates of what has happened, the program’s net economic value since inception now stands at about -$5.4 billion. We think FHA’s current projections are a bit optimistic, but even using their old numbers, the program’s economic value held steady.

The details of the HECM program’s bottom line: probably a little over $5.3 billion in MIP collected to date (in present value terms), minus at least $0.7 billion in realized losses, minus the $7.9 billion in projected negative net present value (“NPV”), which includes nearly all of the $4.3 – $7.9 billion baked-in crossover loss we estimate, minus another $2.1 billion in projected losses for HECMs originated in FY 2009 – 2011. In sum, about $5.4 billion negative NPV for the program since its inception, with the vast majority of the damage coming from the 2005 – 2008 HECM loan cohorts. Even if FHA’s assumptions are too rosy by $2 billion, the program is still holding the bottom line in unfavorable economic conditions.

$5.3 Billion   MIP collected
-$0.7 Billion   Realized losses
-$7.9 Billion   1990-2008 originations: projected net loss
-$2.1 Billion   2009-2011 originations: projected net loss
-$5.4 Billion   HECM program Net Present Value 1990-2011

In our last blog on this topic, we note that unlike the forward mortgage industry, the reverse mortgage industry has already undergone the painful process of reform, including reducing loan-to-value ratios (principal limits) and weaning itself off Fannie Mae. Barring another housing catastrophe, the worst may be over. If so, the current position of the HECM will improve each year, as FHA’s HECM risk profile reflects an increasing percentage of the new, more conservative standard HECM loans and HECM Savers.

FHA Fiscal Year 2011 Annual Reports – Still Too Rosy

Friday, November 18th, 2011

HUD released its FY 2011 annual reports on November 15, 2011, including the HUD Mutual Mortgage Insurance (MMI) report, the FHA Annual Management Report (AMR), and the HECM MMI Actuarial Analysis. Upon review, our conclusion is that HUD is still significantly understating the expected future losses in the HECM book of business. Others have concluded the same about the forward loan book: The Wall Street Journal Editorial page took HUD to task on this point Thursday, November 17, 2011, and Wharton Professor Joseph Gyourko published a study for the American Enterprise Institute’s for Public Policy Research asking “Is FHA the next housing bailout?”

There are three central points addressed in this post:
1. Aggregate changes to the FHA’s overall HECM exposure;
2. Comments on the FHA’s prepayment conclusions; and
3. FHA’s Home Price Appreciation (HPA) assumptions, T&I defaults, and their impact on future losses.

Overall HECM Exposure

It is important to note the FY 2011 HECM Actuarial Analysis covers only the MMI Fund, not the GI Fund, where most of the HECM program’s exposure rests. While the Economic Value of the loans in the MMI Fund may be positive, it represents just 37% of HECMs, all of which were originated in 2009-2011, after the recent real estate bubble deflated. The brunt of the analysis should have focused on the GI fund. Instead, the economics of the GI fund is mentioned briefly deep in the notes to the FHA financial statements in the AMR.

We reported in our April posting that the Net Present Value of the entire HECM program was -$7.3 billion as of 9/30/10. Our calculation took FHA’s FY 2010 projected loss of $11.365 billion and offset it with historical receipts ($4.5 billion), and estimated realized losses ($400 million). The bulk of this $11.365 billion estimate comes from projecting future losses on existing books of business, primarily those loans originated in years 2005-2008, which account for 47% of all HECMs ever originated. This group of HECMs was originated at the height of the home appreciation bubble, and as a result, the overall program is particularly sensitive to prepayment assumptions and HPA forecasts, especially in the later years.

For FY 2011, The GI Fund reported Liabilities for Loan Guaranties (LLG) of $7.864 billion dollars, down $828 million from last fiscal year’s $8.692 billion projection. The LLG is defined as the net present value of anticipated cash outflows and cash inflows. The MMI fund, which has insured HECMs since FY 2009, reported an LLG of $2.149 billion, down $524 million from FY 2010’s $2.673 billion projection. Therefore, FHA’s overall projected exposure to future HECM losses is reduced by $1.352 billion to $10.013 billion.

How is this projection calculated? The two principal drivers for future performance in the HECM book of business are 1) future HPA assumptions; and 2) prepayment assumptions. Because reverse mortgage exposure to loss is back-ended, the magnitude of loss is particularly sensitive to expected home prices in the distant future. Future losses are also sensitive to prepayment speeds because the more slowly the loans pay off, the longer they are outstanding and subject to future home price appreciation/depreciation.

Prepayment Analysis

The Actuarial Analysis “HECM Termination Rates” Exhibit A3.1 (page A-10) suggests an annual prepayment rate of 6% even for new vintages. (The section on prepayments in the Actuarial Analysis is confusing and hard to follow.) Amid the blizzard of statistical formulae, we cannot find anything that resembles recent actual experience. In the AMR for example, we are told 73,093 HECMs were originated in FY 2011. Later in the report, it states that 560,843 HECM loans remained outstanding at the end of the year; the corresponding number in last year’s report was 510,144. That implies that 22,394 loans paid off, a prepayment rate of approximately 4.4%. One can easily do this math going back through previous years’ FHA AMRs to find that HECM prepayment rates have been falling steadily for almost six years. For the largest cohorts, the bubble-year vintages, prepayment rates are currently closer to 2%. See New View Advisors’ Prepayment Index tab here.

New View Advisors’ modeling of prepayment activity also incorporates mortality, mobility, and refinance, but in this environment, the latter two factors equal nearly zero. Actuarial life expectancy tables have been used since 1999 with extreme accuracy. They are a highly predictive indicator of prepayments, and much easier to comprehend. The tables we use are Annuitant mortality tables, not those of the population as a whole; this is essential for capturing the adverse selection of seniors who choose long-term financial products. The FHA Actuarial Review report states “base-case mortality rates were based on the 1999-2001 U.S. Decennial Life Exhibit, published by the Center for Disease Control and Prevention in 2004.” These tables appear to be for the population as a whole, and for major subgroups, but do not address annuitants. This may have introduced a serious sampling error into the Actuarial Review’s prepayment forecast.

Mortality is the dominant driver of Maturity Events, and depressed home prices slow down the rate at which matured HECMs are liquidated (i.e. loan paid off and/or property sold), putting still more downward pressure on prepayment rates. FHA’s faster prepayment assumptions are painting too rosy a picture of future expected loss.

Home Price Appreciation (Depreciation)

As for the impact on future HPA, each of the five scenarios portrayed in the Actuarial Review fails to contemplate the effect of a prolonged real estate slump or the effect of volatility on future losses. As mentioned above, because reverse mortgages suffer realized losses at the end of their life, short-term swings in home prices do not have nearly the deleterious effect on loss that long-term economic trends do. Exhibit B2.1 in Appendix B of the Actuarial Review clearly depicts the five different HPA scenarios used to test sensitivity. Every scenario has HPA returning to between 3% and 5% annual growth by 2015, with a relatively smooth curve that ignores the volatility of the housing market.

Home prices rose at an unprecedented 6% annual growth from HECM’s inception in 1989 until 2007, and HPA was 3-4% from 1945 until the bubble in 2007, but this doesn’t predict future home values. Assuming this period as the norm conveniently forgets that it is bracketed by the Great Depression and the Great Housing Bust. One could argue the recent 30% drop in HPA is the reversion to the mean, and that a relatively quick return to 5%, then 3.4% (Page 6 of the Actuarial Review) annual growth is not a plausible “Base Case.” What would happen if HPA spiked and fell again? That would also be devastating, as a new book of business would be originated at another temporary peak in the real estate market. While no one has a crystal ball, at a minimum, the FHA should be showing the effect of volatility and a prolonged flat or negative HPA in some of its sensitivity scenarios.

If we can’t predict the future, we at least can examine the past. Home prices fell over the last year, at about 5% on average during FY 2010. If prepayments were slower than expected and home prices fell, how can FHA justify an LLG decline of nearly $1.4 billion?

Default

The Actuarial Review Appendix D forecasts a T&I default rate of 2.2%. Once again, this bears no semblance to recent reported experience. At the annual NRMLA conference in Boston last month, the FHA themselves reported T&I defaults in its portfolio were running north of 8% and growing, nearly four times this assumption. Where did 2.2% come from? It is neither descriptive of historical experience nor plausible as a forecast. Unfortunately, the severity of loss for defaulted loans was not disclosed, so it is not possible to understand the impact of default on realized losses for these books of business.

Conclusion

FHA’s $10 billion expected loss from its HECM book is likely low, based on overly optimistic HPA, loss, and prepayment assumptions. How far off is anyone’s guess, but sensitivity analysis is supposed to show worst case, base case, and best case execution, and these reports have failed to do so. Nonetheless, we emphasize that FHA has taken many of the proper steps necessary to reform the HECM program. The reverse mortgage industry has reduced the amount it lends, weaned itself off Fannie Mae, and introduced financial assessment to minimize loss.
Considering HECM is less than 1% of the overall mortgage market, the FHA needs to take a sober look at the forward side of its business. The astonishingly high Loan-to-Value ratios of FHA’s forward mortgage programs are a good place to start.

Understanding Reverse Mortgage Prepayments: Focus on Seasoned Reverse Mortgage Loans

Thursday, October 21st, 2010

Part II: Loan Age vs. Borrower Age

No one takes out a home mortgage loan with the intention of repaying it immediately. A mortgage loan is a long-term loan designed to finance a long-term asset. Some mortgages do payoff quickly, mostly due to unexpected life events or refinancing opportunities. However, prepayments in the first months after mortgage origination tend to be significantly lower than seasoned mortgages, which are often defined as mortgages that have been outstanding for more than two years.

When analyzing mortgage prepayments, the investor should take this into account. It is useful to segregate prepayment rates into seasoned and unseasoned loans, because otherwise the calculated prepayment rates may be artificially low. Also, without controlling for loan seasoning, it becomes more difficult to isolate the impact of other variables, such as borrower age.

Loan Age (or Seasoning) is a well-researched subject for forward mortgages, but for reverse mortgages there is another age that matters even more: the age of the borrower. Borrower Age is a strong driver of prepayment rates, and unlike Loan Age, its importance increases as time passes. FHA’s recent release of all HECM prepayment history from program inception through January 2010, allows for a comprehensive look at prepayment behavior by age cohorts over time. If we look at the borrower age (or age of the youngest borrower, in the case of multiple borrowers), a strong correlation emerges.

For seasoned HECMs, 64 year-old borrowers have prepaid at a rate of 4.3% per year, 74 year-olds at a rate of 6.7% per year, and 84 year-olds at 10.1%. Prepayments for reverse mortgages are the result of loans paid off from borrower Maturity Events. Each of these prepayment percentages is the equivalent of a fraction, the numerator of which is the number of loan payoffs (to the investor) corresponding to the borrower age at the time of payoff, and the denominator of which is equal to the outstanding number of loans corresponding to that borrower age. For example, the 74 year old cohort represents all HECM loans originated to 72-year-old borrowers that were still outstanding after 2 years, plus all HECM loans originated to 71-year-old borrowers that were still outstanding after 3 years, and so on.

The correlation is strongly linear, with each year of borrower age resulting in an approximate 0.3% increase in prepayment speed. This corresponds very closely to the average increase in mortality rates by age cohort in the senior citizen population of the U.S., especially those 70 to 80 years old, which represent the bulk of the observations in our prepayment data.

Of course, there are other factors that affect HECM prepayments. We have included a comprehensive history of HECM and proprietary RM loans side by side on this spreadsheet. The effect of good and bad economic times can be seen, most dramatically during the recent recession, but also during the previous recession in 2001. It is also evident that proprietary reverse mortgage borrowers generally pay much faster than HECMs, likely due to their higher mobility and ability to refinance into other mortgage products.

The Borrower Age correlation is perhaps the most important concept in reverse mortgage prepayments, and a source of value as well. Although most mortgage investors are not accustomed to dealing with actuarial risk, it is worth spending time to understand how the actuarial nature of reverse mortgages reduces prepayment risk.

Understanding Reverse Mortgage Prepayments: Focus on HECMs

Monday, July 26th, 2010

Part I: Mortgage Prepayment Risk in Reverse

Mortgage investors spend a great deal of time and effort analyzing prepayment risk, and with good reason. The mortgage borrower usually receives a valuable option: the ability to prepay at any time in whole, or in part, without penalty. This is sometimes easier said than done, especially in the current economic climate, but nonetheless the mortgage investor is short this option. Mortgage borrowers refinance to reduce their monthly payment and/or to cash out home equity that, under normal conditions, rises as the loan balance decreases, and the home value increases. Compounding the investor’s uncertainty is the fact that prepayments rise when interest rates fall (when reinvestment rates are lower), and prepayments fall when interest rates rise (when reinvestment rates are higher). As a result, mortgage investors must analyze the key drivers of forward mortgage prepayment rates: the mortgage interest rate, current mortgage rates, loan seasoning, the volatility of interest rates, and the available equity and refinancability of the underlying property.

For Reverse Mortgages (RM), the prepayment picture is fundamentally different. Borrowers don’t make a monthly mortgage payment, so their refinancing incentive is diminished. Under normal conditions, the RM borrower’s home equity decreases, further diminishing the attractiveness of refinancing. The age of the borrower matters more than the age of the loan. These are all natural consequences of the characteristics of the reverse mortgage loan, a product that combines mortgage and actuarial-based finance to provide liquidity to the senior homeowner, without the burden of a monthly mortgage payment.

As a result, reverse mortgages provide the investor with a unique opportunity: a mortgage with considerably less market-driven prepayment risk, governed instead by predictable actuarial factors, capable of sustaining premium value over a sustained period. For HECM/HMBS investors, FHA provides an additional benefit: the 98% Maximum Claim Amount “put” that caps extension risk and shortens the duration of their investment in a predictable, measurable way. Informed mortgage investors have come to appreciate the reverse mortgage prepayment advantage, which goes a long way in explaining the recent HMBS boom.

But how do we measure reverse mortgage prepayments? By which yardstick? Using what data? Doesn’t refinancing sometimes happen, into a new reverse mortgage, or another kind of mortgage? Also, there haven’t been “normal conditions” in the housing market for some time. Instead, a period of rapid home depreciation has followed a period of rapid appreciation, throwing the housing market into turmoil. These are all important questions, but we can’t answer all of them in one essay. Suffice to say, the study of mortgage prepayments is practically a science unto itself, so let’s keep it simple (at least for Part I) and understand the basics.

Finding The Data
New View Advisors began publishing a prepayment index for seasoned HECMs and proprietary RMs several months ago. We compile this data from our clients and other private sources. The prepayment rate is simply a fraction, expressed as a percentage, the numerator of which is the number of loans paid in a given period of time, and the denominator of which is the number of loans outstanding at the beginning of the same period. Importantly, we define HECM prepayments from the investor’s perspective, which means the earlier of the loan prepayment or assignment to FHA.

The most commonly used measure is Constant (or Conditional) Prepayment Rate, or “CPR,” equal to annualized prepayment rate over a stated period. For example “1 Month CPR” is equal to the annual rate of prepayment, based on one month of prepayments. (The Monthly Rate is referred to as “Single Monthly Mortality,” or SMM, and is approximately equal to one-twelfth of the 1 Month CPR.) 12-Month CPR is a useful yardstick as it averages out a year of data, smoothing out seasonal and other anomalies.

For seasoned HECMs, our index showed 12-Month CPRs of 4.4%, 4.6%, 4.7%, 4.5%, and 4.2%, for the 12 months ending September 2009, October 2009, November 2009, December 2009, and January 2010, respectively. Proprietary loans generally pay faster (in forward and reverse mortgages), and recent experience is no exception.

FHA published a very useful HECM prepayment study in 2007, but its relevance diminished as time passed, especially as that time reflected very different economic conditions. Also, this study presented the data in summary format; no loan level data was made available to the public.

However, FHA recently published a mother lode of HECM prepayment data, a comprehensive loan level file disclosing the main loan characteristics for every HECM loan that ever closed (and a number that didn’t) from the inception of the program through January 2010. The FHA plans to update this file every six months.

Analyzing the Data
The new FHA data reveals a great deal. Since 2007, as the New View Advisors index also shows, prepayments have fallen dramatically. Under normal conditions, they should rise for each vintage of origination, as borrowers age and an increasing percentage of loans are put back to HUD at the 98% Maximum Claim Amount.

A comparison of the 1999 vintage and the 2006-2007 vintages illustrates the impact of the housing crisis, as seen on this spreadsheet. During the unseasoned period (i.e., the first two years) both cohorts show very low prepayments. In fact, they both pay at 2.79% CPR within the first three months after their origination period. But after that, their paths diverge. Aided by rising home values, and presumably much higher homeowner mobility, the 1999 vintage climbs to a double-digit CPR in a little over two years. But the 2006-2007 vintage has never approached double digits, and has yet to crack even 5% CPR over a 12 month period. Interestingly, reverse mortgage prepayments have been falling even during a period of record low interest rates — hardly in line with conventional mortgage prepayment behavior.

Nonetheless, prepayments for this cohort are rising ever so slowly. The 2006-2007 vintage roughly approximates the population used for our index, and indeed confirms our published numbers, averaging a mere 0.2% difference.

Breaking down the numbers by borrower age, we see a strong correlation, with prepayment rates rising from 3.5% for 62 year olds, 6.9% for 77 year olds, and 26.1% for borrowers 90 and older. These numbers reflect HECM experience since the program’s inception; current prepayment numbers are lower across the board, but still show a strong borrower age correlation.

Conclusions
We will have much more to say about this topic, but for now the data confirms that reverse mortgages have unique prepayment characteristics that derive from the unique features of those mortgages and their unique borrower population. However, they are hardly immune from prevailing economic conditions. HECM prepayment rates have fallen dramatically as the economy has decimated mobility and refinancing opportunities. Also, the FHA assignment feature significantly protects investors against extension risk, and prepayments for a given cohort will still tend to rise over time, however slowly.

The strength of the borrower age/prepayment correlation commands investor attention, not only in analyzing existing loans and HBMS pools, but also with respect to the future of tailor-made HMBS pools, comprised of narrow ranges of borrower age groups (and therefore tailor-made ranges of likely duration.)

The new FHA data are an important contribution to the understanding of HECM prepayments. Supplemented with the monthly updates of our own index, they help the proper analytic foundation for understanding the risks and rewards of reverse mortgage investments.

For more detailed information on prepayment analysis by vintage, borrower age, loan age, HECM type, etc., please contact us.