The Federal Reserve Board eagle logo links to home page

Skip to: [Printable Version (PDF)] [Bibliography] [Footnotes]
Finance and Economics Discussion Series: 2009-42 Screen Reader version

Reversing the Trend: The Recent Expansion of the Reverse Mortgage Market

Hui Shan*

Keywords: Reverse mortgages, housing, aging

Abstract:

Reverse mortgages allow elderly homeowners to tap into their housing wealth without having to sell or move out of their homes. However, very few eligible homeowners have used reverse mortgages to achieve consumption smoothing until recently when the reverse mortgage market in the United States witnessed substantial growth. This paper examines 1989-2007 loan-level reverse mortgage data and presents a number of findings. First, I show that recent reverse mortgage borrowers are significantly different from earlier borrowers in many respects. Second, I find that borrowers who take the line-of-credit payment plan, single male borrowers, and borrowers with higher house values exit their homes sooner than other reverse mortgage borrowers. Third, I combine the reverse mortgage data with county-level house price data to show that elderly homeowners are more likely to purchase reverse mortgages when the local housing market is at its peak. This finding suggests that the 2000-05 housing market boom may be partially responsible for the rapid growth of reverse mortgage markets. Lastly, I show that the Federal Housing Administration (FHA) mortgage limits, which cap the amount of housing wealth that an eligible homeowner can borrow against, have no effect on the demand for reverse mortgages. The findings have important implications to both policy-making and the economics of housing and aging.


1 Introduction

Housing wealth is often the largest non-pension wealth component for many elderly homeowners. For example, the 2007 Survey of Consumer Finances (SCF) data suggest that for 6.5 million homeowners aged 62 or above, housing wealth represents at least 80% of their total wealth. To determine whether Americans are saving enough for retirement, it is crucial to decide how housing wealth should be treated: Are elderly homeowners willing and able to use their housing equity to finance spending in retirement? According to the 2007 SCF, 4.2 million homeowners aged 62 or above had a house-value-to-income ratio of at least 10. For these house-rich elderly homeowners, reverse mortgages would seem to carry the potential to increase their consumption while allowing them to continue living in their homes.

The most common type of reverse mortgage is the Home Equity Conversion Mortgage (HECM), insured by the Federal Housing Administration (FHA) and constituting over 90% of the U.S. reverse mortgage market.1 Despite its potential economic appeal, using reverse mortgages to finance consumption after retirement has been the exception rather than the rule among elderly homeowners. From its inception in 1989 to the end of 2007, out of tens of millions of eligible homeowners, less than 400,000 loans were originated through the HECM program.

Although usage remains relatively low, reverse mortgage markets have grown substantially in recent years. For example, in the early 1990s, only a few hundred HECM loans were originated each year. In contrast, in 2007, over 100,000 reverse mortgage loans were originated through the HECM program. Plausible explanations include rising house values, lower interest rates, and increasing awareness of the product. Also, whether the expansion is transitory or permanent remains unclear. Addressing these questions is not only essential to understanding elderly homeowners' desire to consume housing wealth, but it also provides evidence for policymakers to conduct the cost-benefit analysis and to design more efficient policies in facilitating the conversion of home equity into liquid assets.

In this paper, I examine all HECM loans that were originated from 1989 through 2007. I first present descriptive evidence on the characteristics of HECM borrowers and HECM loans with an emphasis on how these characteristics have changed over time. I then study the loan termination and assignment outcomes using probit and proportional hazard models. A HECM loan is terminated when the borrower dies or permanently moves out of the house. The estimates suggest that borrowers who choose the line-of-credit payment plan, single male borrowers, and borrowers with more expensive homes terminate their HECM loans sooner than other reverse mortgage borrowers. In addition, I investigate the link between house price appreciation and the growth of reverse mortgage markets. I show evidence suggesting that areas at the peak of the housing market tend to experience more increases in HECM loan origination. Lastly, I examine the role of FHA mortgage limits in the reverse mortgage market. The county-specific FHA mortgage limits effectively cap the amount of housing equity a borrower can convert into liquid assets. I find no evidence to support the claim that FHA mortgage limits have prevented reverse mortgages from becoming more popular.

To the best of my knowledge, this is the first paper using 18 years of HECM loan-level data to systematically study the reverse mortgage market. I take the first step to address questions such as what factors have contributed to the rapid growth of reverse mortgage markets in recent years. According to the Census Bureau, the number of persons above age 65 will increase to 40 million in 2010 and increase further to 81 million in 2040. The questions addressed in this paper will become increasingly important with the population aging and the baby-boomers' entering retirement age.

The rest of this paper proceeds as follows. Section 2 overviews the existing literature on reverse mortgages and introduces the HECM program in detail. In section 3, I describe the data used in this paper. I then show the characteristics of HECM loans and HECM borrowers in section 4. Sections 5, 6, and 7 discuss the empirical evidence on loan termination and assignment outcomes, house price appreciation, and FHA mortgage limits, respectively. The last section concludes and clarifies common misconceptions about the reverse mortgage market.

2 Background

2.1 Overview of Existing Studies

The question of whether Americans are financially prepared for retirement has inspired heated debates in the literature.2 When evaluating retirement saving adequacy, economists and financial planners have to decide whether housing equity should be included as consumable wealth. Because housing is both a consumption good and an investment good, the correct treatment of housing equity may not be obvious in the retirement saving context. For example, while Mitchell and Moore (1998) add housing equity to household net worth, Bernheim et al. (2000) exclude it in their calculation. More recently, Sinai and Souleles (2008) suggest that the fraction of "consumable housing equity" ranges from 60% to 99% for elderly homeowners depending on their age.

To what degree we should consider housing equity as retirement savings depends on to what degree elderly homeowners are willing and able to consume their housing wealth. It is well known that many seniors prefer staying in their homes for as long as they can. For example, in a survey sponsored by the American Association of Retired Persons (AARP), 95% of persons 75 and older agreed with the statement "What I'd really like to do is stay in my current residence as long as possible."3 A series of studies by Venti and Wise (e.g., 1989, 1990, and 2004) show that elderly homeowners do not reduce their housing wealth in the absence of precipitating events such as the death of a spouse or a move to a nursing home. If elderly homeowners have strong psychological attachments to their homes, then reverse mortgages, which generate additional income and liquid assets for elderly homeowners while allowing them to continue living in their homes, may be welfare-improving for many households.

A number of studies have estimated the potential size of the reverse mortgage market. Venti and Wise (1991) analyze the 1984 Survey of Income and Program Participation (SIPP) data and find that a reverse mortgage in the form of annuity payments would substantially affect the income of the single elderly who are very old. Merrill et al. (1994) use the 1989 American Housing Survey (AHS) data to show that out of the 12 million elderly homeowners who own their homes free and clear, 800,000 could benefit substantially from reverse mortgages. Mayer and Simons (1994) consider both income increases and debt reductions as benefits of reverse mortgages and find a much larger potential market for reverse mortgages than previous studies: Over 6 million homeowners in the U.S. could see their effective monthly income being raised at least 20% by reverse mortgages.

In practice, the reverse mortgage market is much smaller than expected. On the demand side, a number of factors may have prevented the reverse mortgage market from growing bigger. First, elderly homeowners with strong bequest motives may not find reverse mortgages attractive because reverse mortgages reduce the amount of wealth they can leave to their estates. Second, the probability of shouldering large medical expenses increases over time for the elderly. In the absence of other forms of protections such as Long-Term Care Insurance, many elderly homeowners use their housing equity to self-insure. Using a survey conducted on 2,673 homeowners aged 50 to 65, Munnell et al. (2007) report that nearly one-half of the respondents who claim they are not planning to tap their housing equity in retirement list "insurance against living and health expenses" as the reason. Davidoff (2008a) presents a model suggesting that such behaviors may even be optimal. Third, certain features of the HECM program and its interaction with some welfare programs may be undesirable. For example, a HECM loan usually requires large upfront costs, the amount of home equity against which one can borrow is capped by the FHA mortgage limits, and the additional income received from a HECM loan may disqualify one from public assistance such as Supplemental Security Income (SSI) or Medicaid. Michelangeli (2008) argues that reverse mortgages could be welfare-reducing if borrowers face significant moving risks. Fourth, reverse mortgages are complex financial products and can be particularly challenging for elderly homeowners. Conversations with people in the industry suggest that many senior homeowners have misconceptions about reverse mortgages. Lastly, the elderly may value owning their homes free and clear so much that they are averse to the idea of borrowing against them.

On the supply side, lenders face various obstacles as well. First, reverse mortgages are significantly different from traditional "forward" mortgages. Lenders with little experience in the reverse mortgage market often have to confront unfamiliar documentation and regulatory requirements. For example, lenders who are accustomed to forward mortgages have to prepare different documents for reverse mortgages to satisfy the Truth-in-Lending Act requirements. As a result, lenders must designate reverse mortgage specialists among their employees. Because the HECM program caps origination fees charged by lenders, such a move is only economical if there is a sufficient volume of HECM loan originations. Another consideration is that different states have different regulations with respect to reverse mortgages. To comply with such regulations, lenders who operate in multiple states have to bear additional costs. In addition, because of the unconventional cash-flow pattern, reverse mortgages are difficult to securitize and finance. In fact, according to Szymanoski et al. (2007), HECM loans were not securitized until August 2006. Finally, the Fair Housing Act prohibits pricing loans based on gender, despite the fact that elderly males and females have very different life expectancies.

Besides the factors discussed above, economists have also recognized that reverse mortgage markets may suffer from adverse selection and moral hazard problems. Because reverse mortgage loans are not due until the borrower dies or permanently moves out, people who know they are likely to stay in their homes for a long time will find reverse mortgages more attractive than others. However, Davidoff and Welke (2007) find advantageous selection in the HECM program. In other words, HECM borrowers appear to exit their homes at a faster pace than the general population. The authors suggest that higher discount rate among the borrowers combined with house price appreciation may explain observed advantageous selection. Furthermore, economists are concerned that the moral hazard problem of home maintenance would make lenders think twice before entering the reverse mortgage market. Davidoff (2006) uses AHS data to show that homeowners over 75 years of age spend less on routine maintenance than younger owners of similar homes.

Overall, most of the studies on reverse mortgages do not have loan-level data and, therefore, have to rely on hypothetical borrowers. Among the few studies that do look at loan-level data, Davidoff (2006) and Szymanoski et al. (2007) focus only on termination rates of HECM loans, and Case and Schnare (1994) and Rodda et al. (2000) analyze only the data from earlier years of the HECM program. Given that 88% of all HECM loans originated from 1989 through 2007 were taken out after 2000, the field calls for research using more recent data. This paper aims to fill the gap.

2.2 Background on the HECM Program

The Congress established the HECM program in 1987 and authorized the Department of Housing and Urban Development (HUD) to administer the program. The first HECM loan was made in 1989. Since then, the HECM program has been the dominant reverse mortgage product in the United States.4 To be eligible for a HECM loan, borrowers have to be 62 years of age or older, and occupy the property as their principal residence. Borrowers must either own their homes outright or have a small mortgage balance. A HECM loan is a "non-recourse" loan, meaning that the borrower and her estate will never be required to pay more than the value of the property and that no other assets can be seized to repay the loan.

HECM loans differ from traditional home equity loans or home equity line of credit (HELOC) in two ways. First, a HECM loan does not have a fixed maturity date. The loan becomes due and payable only after the borrower dies or the borrower no longer occupies the property as a principal residence. Second, while home equity loans and HELOCs generally require borrowers to have sufficient income and creditworthiness, HECM loans do not have such underwriting requirements.5 Therefore, house-rich but cash-poor elderly homeowners who cannot obtain home equity loans may find HECM loans particularly attractive.

The amount that the borrower can receive from a HECM loan is calculated in three steps. The first step is to determine the Maximum Claim Amount (MCA) - the lesser of the appraised value of the property or the county-specific FHA mortgage limits for a one-family residence under section 203 (b) of the National Housing Act. The limits are usually set at 95% of the median sales prices for any given county in a given year, although MCAs are subject to both ceiling and floor levels, creating nationwide maximum and minimum values for the FHA mortgage limits. For example, the ceiling was $362,790 and the floor was $200,160 in 2007.

The second step is to determine the Initial Principal Limit (IPL) by multiplying the MCA by a factor between zero and one. The magnitude of the factor depends on the age of the borrower and the "expected interest rate" at the time of loan closing.6 The expected interest rate, a proxy for future interest rate, equals the sum of the ten-year Treasury rate and the lender's margin. The lender's margin is typically between 100 and 200 basis points. These principal limit factors are designed such that, under certain assumptions, the loan balance reaches the MCA at the time when the loan becomes due in expectation.7 As a result, the factor increases with the borrower's age and decreases with the expected interest rate. For example, the factor equals 0.281 for a 65-year-old at a 10 percent expected rate, and it equals 0.819 for an 85-year-old at a 5 percent expected rate.

The third step is to calculate the Net Principal Limit (NPL), which is the amount the borrower can take as a lump sum in cash at closing, by subtracting from the initial principal limit the upfront costs associated with HECM loans and a set-aside for a monthly servicing fee. The upfront costs include the initial Mortgage Insurance Premium (MIP), origination fee, and closing costs. The initial MIP is set at 2% of the MCA. The origination fee is set at $2,000 or 2% of the MCA, whichever is greater.8 Closing costs include origination fees and other third-party fees such as appraisal fees, credit report fees, and title insurance fees. Overall, the upfront costs on a HECM loan range between $7,000 and $20,000. A servicing fee set-aside is the present value of the monthly servicing fee charged by the lender. The typical monthly servicing fee is $30 or $35. Both the upfront costs and servicing fees are financed rather than paid by the borrower out of pocket. Figure 1 summarizes the steps described above in calculating the net principal limit.

Given the amount of net principal limit, HECM borrowers can choose from five payment plans to receive the mortgage proceeds. Under the Tenure plan, the borrower will receive equal monthly payments from the lender for as long as the borrower lives and continues to occupy the property as her principal residence. This payment plan is also called a "reverse annuity mortgage" in the literature due to its resemblance to an annuity product. Under the Term plan, the borrower will receive equal monthly payments from the lender for a fixed period of months selected by the borrower. Note that even though payments stop at the end of the selected term, the loan is not due until the borrower dies or moves out of his or her home. Under the Line of Credit plan, the borrower will receive the mortgage proceeds in unscheduled payments or in installments, at times and in amounts of the borrower's choosing, until the line of credit is exhausted. This payment plan is the most popular among HECM borrowers. In addition, the Modified Tenure plan allows the borrower to combine a line of credit with a tenure plan. The Modified Term plan allows the borrower to combine a line of credit with a term plan.9 Table 1 shows the principal limit factor, net principal limit, the monthly payment under a tenure plan, and monthly payment under a 10-year term plan for a hypothetical borrower, assuming a MCA of $200,000, an initial MIP of $4,000, an origination fee of $4,000, closing costs of $2,000, and a monthly servicing fee of $30.

One key feature of HECM loans is the FHA insurance program. Under this program, HUD insures the borrower against the risk that the lender can no longer make the contracted payments. It also insures the lender against the risk that the loan balance exceeds the property value. For example, lenders can assign loans to HUD when the loan balance reaches 98% of the MCA. In the event that the proceeds from the sale of the property are not sufficient to pay the outstanding loan balance, lenders who have not assigned the mortgage to HUD can submit a claim for insurance benefits up to the MCA. To pay for this insurance program, HUD charges a Mortgage Insurance Premium (MIP). The initial MIP, as mentioned before, is set at 2% of the MCA. The monthly MIP is set at an annual rate of 0.5% and is charged on the outstanding balance of a HECM loan.

3 Data Description

The data I analyze in this paper are the loan-level HECM data provided by HUD. I have information for all HECM loans made between 1989 and 2007, a total of 387,999 records in the raw data, including the age of the borrower, age of the co-borrower, gender and marital status of the borrower, the appraised value of the property at origination, location of the property (i.e., state and ZIP code), the Maximum Claim Amount (MCA), expected interest rate, Initial Principal Limit, choice of payment plan, monthly payment amount, loan origination date, loan termination date, loan assignment date, whether a claim was filed to HUD by the lender, and the nature of the claim.

In the process of data cleaning, I dropped irregular observations such as loans with missing borrower age, missing gender or marital status indictor, and loans that were not endorsed by HUD for various reasons. I also dropped borrowers from places other than the continental United States because the FHA mortgage limits for Alaska, Guam, Hawaii and the Virgin Islands are very different from the rest of the country. After these procedures, I have a final sample of 375,392 observations.

Figure 2 shows the number of HECM loans originated each year beginning 1989 through 2007. In the early years of the program, only a small number of elderly homeowners took out HECM mortgages. In contrast, loan origination has grown substantially in recent years. For instance, the number of loans made in 2007 is ten times the number in 2001. One of the potential explanations for such significant growth is that elderly homeowners have become more comfortable taking equity out of their homes and taking on debt in general. To test the plausibility of this explanation, I used the 1989-2007 SCF data and plotted in Figure 3 the fraction of homeowners aged 62 or above who have credit card debt, debt secured by their primary residences, or any type of debt. The fraction of elderly homeowners with credit card debt trended up slightly over the 18-year sample period. The fraction of elderly homeowners with mortgages, home equity loans, or home equity lines of credit increased steadily from 22% in 1989 to 39% in 2007. The fraction of elderly homeowners with any type of debt rose from 44% to 57%. The SCF data suggest that an increasing proportion of elderly homeowners feel comfortable taking on debt secured by their homes. However, such a shift in financial attitude is unlikely to fully explain the unusually large increase in reverse mortgage origination. Other factors may also have contributed to the growth of the HECM program, including the 2000-05 housing market boom, lower interest rates in recent years, and more awareness of reverse mortgages among elderly homeowners.

To perform the empirical analysis shown later in this paper, I merged the HECM loan-level data with a number of public and proprietary data sets. First, I used the 2000 Census ZIP code level data to characterize the neighborhoods where reverse mortgages were originated. Second, I used the monthly county-level house price indexes purchased from First American CoreLogic to study the relationship between house price appreciation and reverse mortgage market growth. This data set covers 679 counties in the United States and merged with approximately 90% of the loans in the HECM data set. Third, I used the 1990-2008 county-level FHA mortgage limit data to examine the effect of these limits on the demand for reverse mortgages. Lastly, I used a data set purchased from the United States Postal Service to match ZIP codes with counties for each HECM loan in my sample.

The administrative data analyzed in this paper are essential for studying the reverse mortgage market, because reverse mortgage borrowers are a tiny fraction of the general population and they are unlikely to be captured by public surveys. In addition, administrative data tend to be more accurate than self-reported data in most public surveys. Nevertheless, there are a couple of caveats associated with the data. First, similar to many administrative data sets, we do not know very much about these borrowers beyond their characteristics that are used in the HECM pricing model. For example, we do not know the income and financial wealth of these borrowers, nor do we know their demographic characteristics such as race, eduction, and number of children. Second, according to the staff member at HUD who shared the data with us, our data come from a snapshot at the end of 2007. Because borrowers are allowed to change their payment plans at a small cost, it is possible that the payment plan we observe is not the original payment plan chosen by the borrower.10 Moreover, if borrowers chose the line-of-credit payment plan, I do not observe when and by how much they drew down the line of credit.

4 Characteristics of HECM Borrowers and Loans

In this section, I first compare reverse mortgage borrowers with the general population. I then show the difference between borrowers who took out reverse mortgages in the early years of the HECM program and borrowers who took out reverse mortgages in 2007. This comparison highlights the significant changes in HECM borrower and loan characteristics over time. In addition, I present evidence on where reverse mortgage borrowers come from, using the geographic identifiers in the HECM data.

Reverse mortgage borrowers are generally older than eligible homeowners in the general population. The median age of HECM borrowers at the time of loan origination is 73.5, and the median age of homeowners 62 years or older in the 2000 Census is 72.0. Single males and single females are more likely to take out reverse mortgages than married couples. According to the 2000 Census, 52.8% of homeowners aged 62 or above are married couples. However, only 36.0% of reverse mortgage borrowers are married couples. Moreover, reverse mortgage borrowers tend to have more expensive houses than the general population of elderly homeowners. Panel B of Table 2 compares the median home value of HECM borrowers with that of the SCF respondents who are homeowners aged 62 or older.

Characteristics of HECM borrowers and HECM loans change substantially over time. Figure 4 shows the age distribution of early borrowers and that of recent borrowers; early borrowers refer to loans originated between 1989 and 1999, and recent borrowers refer to loans originated in 2007. Two features of this figure are worth mentioning. First, the distribution of borrower age shifts to the left over time, meaning that recent borrowers are younger than early borrowers at the time of loan origination. Second, there is a spike at age 62 in the histogram for recent borrowers but not in the histogram for early borrowers. Such a spike suggests that homeowners younger than age 62 may want to purchase reverse mortgages if allowed. These two features imply that the demand for reverse mortgages has been growing most rapidly among younger elderly homeowners.

Besides age, Panel A of Table 2 also shows the difference between early borrowers and recent borrowers along other dimensions. For example, an increasing number of single males and married couples entered the reverse mortgage market relative to single females. The average expected interest rate faced by HECM borrowers, which is the ten-year Treasury rate plus the lender's margin, declined from 7.5% in early years to 5.7% in 2007. The fraction of borrowers choosing the line-of-credit payment plan increased from 71.4% between 1989 and 1999 to 86.9% in 2007.

An elderly homeowner who wants to tap into her housing wealth generally has two options: selling the house or taking out a HECM loan, both of which involves considerable transaction costs. While selling the house would allow the elderly homeowner to access all of her housing equity, the FHA mortgage limits may limit the fraction of housing equity she wants to tap into. It has been argued in the literature that reverse mortgages may not be attractive to potential borrowers because of such limitations on the fraction of housing equity against which one can borrow. In Figure 5, I plotted the distribution of the ratio of Initial Principal Limit (IPL) to house value for early borrowers and recent borrowers respectively. Recall that the IPL represents the present value of all payments that may be received by the borrower plus the upfront costs. The higher the IPL-to-house-value ratio is, the larger the fraction of the illiquid housing equity the borrower can access at the time of origination. The IPL is the product of MCA and a factor that increases with age and decreases with expected interest rates. For early borrowers, the average IPL is 54.9% of the house value. For recent borrowers, the average IPL-to-house-value ratio is 65.9%. The increase in the IPL-to-house-value ratio is presumably driven by lower interest rates in recent years.

The HECM data used in this paper have information on state and ZIP code of the property for each loan. Figure 6 is a state map that displays the ratio of "reverse mortgages originated from 1989 through 2007" to "owner-occupied housing units with householders aged 60 or above". The housing units statistics are from the 2000 Census, and the ratio is expressed in percentage terms. Reverse mortgages are most concentrated in District of Columbia, Nevada, California, Colorado, and Utah where the ratio is over 3%. They are least concentrated in Mississippi, West Virginia, North Dakota, Kentucky, Alabama, and Iowa where the ratio is less than 0.5%.

To investigate where reverse mortgages come from at a finer level, I calculated the ratio of "reverse mortgages originated" to "housing units with householders aged 60 or above" for each ZIP code in the HECM sample. I then merged in the 2000 Census ZIP code level demographic and social-economic data and estimate a regression model in which each observation is a ZIP code.

\displaystyle \left( \frac{\text{Reverse Mortgages}}{\text{Housing Units with Elderly Owners}}\right)_{z} = \alpha + \mathbf{\beta} \mathbf{X}_z + \epsilon_z (1)

A vector of demographic and social-economic variables are controlled for, including education, median household income, median house price, race/ethnicity, and the fraction of elderly homeowners in the ZIP code. Column (1) includes all ZIP codes in the 2000 Census that have reasonable values of the control variables.11 In column (2), I dropped non-MSA ZIP codes and control for MSA fixed effects. Comparing the estimation results in column (1) with those in column (2) helps us understand the cross-MSA variation and the within-MSA variation in the geographic distribution of reverse mortgages. In both columns, the dependent variable is constructed using all HECM loans originated between 1989 and 2007. On average, the ratio of reverse mortgages to housing units is 1.11% among all ZIP codes and 1.66% among MSA ZIP codes. Note that the magnitudes of the estimates in the two columns are not directly comparable, because the means of the dependent variable are different.

When all ZIP codes are included, the signs of the estimates suggest that reverse mortgages are more likely to originate in ZIP codes with better-educated residents, lower median incomes, higher median house values, higher fractions of minorities, and lower fractions of elderly homeowners. In contrast, when only the within-MSA variation is used, the estimation results in column (2) show that median income and median house value of the ZIP codes do not correlate with reverse mortgage origination. This difference suggests that reverse mortgages are more likely to originate in income-poor but housing-rich MSAs, but reverse mortgage origination does not differ across ZIP codes with different median income and house value within a MSA.

In column (3) and (4), the dependent variable is constructed using 1989-2002 loans and 2003-07 loans, respectively. I limited the sample to MSA ZIP codes and control for MSA fixed effects. Given the rapid growth in the reverse mortgage market in recent years, it is not surprising that the mean of the dependent variable in column (3) is much lower than that in column (4). Comparing column (3) with column (4) highlights the difference over time in the geographic distribution of reverse mortgages. Between 1989 and 2002, reverse mortgages were more likely to originate in low-income ZIP codes with any given MSA. Between 2003 and 2007, reverse mortgages were equally likely to originate in high-income ZIP codes. Such a shift is consistent with the notion that recent reverse mortgage borrowers may be very different from their earlier counterparts.

In summary, the above analyses indicate that reverse mortgage borrowers are not representative of elderly homeowners in the general population. The characteristics of reverse mortgage borrowers and loans changed notably from the early years of the HECM program to now. Thus, conclusions drawn by studies using early HECM loan data may no longer apply to recent borrowers. Moreover, there appears to be significant heterogeneity in the distribution of reverse mortgages across geographic areas.

5 Termination and Assignment Outcomes

A HECM loan is terminated when the borrower dies or permanently moves out the house. A HECM loan is assigned by the lender to HUD when the loan balance reaches 98% of the MCA. In this section, I examine whether various borrower and loan characteristics are correlated with the termination and assignment outcomes of HECM loans.

Figure 7 compares the termination rates of HECM borrowers with the mortality rates observed in the general population for males and females respectively. Throughout the age distribution, termination rates of HECM borrowers are consistently higher than mortality rates of the general population. For example, the termination rate of a single female HECM borrower at age 75 is 0.093, whereas the female mortality rate at age 75 is only 0.028. Presumably, this difference reflects the residential mobility rate among reverse mortgage borrowers because HECM loans are terminated when borrowers move out of their homes permanently. Potential differences in mortality rates between HECM borrowers and the general population may also contribute to the observed pattern. For example, if elderly homeowners who experienced negative health shocks purchase reverse mortgages to cover medical expenses, then reverse mortgage borrowers are also likely to have a higher mortality rate than non-borrowers. The HECM program assumes that termination rates among borrowers are 1.3 times the age-specific female mortality rate. As shown in Figure 7, this assumption is overly conservative, and HECM borrowers exit homes as a much faster pace.

To understand how loans of different characteristics terminate differently, I first estimated a probit model of the 5-year loan termination outcome.

Pr\displaystyle (Terminate_{iat})=\Phi (\alpha \displaystyle + \displaystyle \beta_1 Term_{iat} + \beta_2 Tenure_{iat} + \gamma_1 Female_{iat} + \gamma_2 Couple_{iat}  
  \displaystyle + \displaystyle \delta log(HV_{iat}) + \eta_a + \theta_t)  

 Terminate_{iat} equals one if loan  i terminates within 5 years of origination.  HV_{iat} is the real house value at the time of loan origination.  \eta_a is a full set of age fixed effects, and  \theta_t is a full set of origination year fixed effects.

Column (1) of Table 4 displays the estimation results. Compared to borrowers who choose the line-of-credit payment plan, borrowers who choose the tenure payment plan are, on average, 13.6% less likely to exit homes permanently within five years of loan origination. Because the line-of-credit option allows borrowers to withdraw home equity more quickly than the tenure option, this finding is consistent with the conjecture that borrowers who believe that they will stay in their homes for a long time tend to choose the more back-loaded payment plan - the tenure option. Compared to single male borrowers, single females are on average 19.0% less likely to exit homes within 5 years. This difference is unsurprising since females tend to live longer than males. Recall that for married couples, reverse mortgages do not become repayable until both spouses die or move out of the house. Married reverse mortgage borrowers are 36.1% less likely to terminate reverse mortgage loans within five years of loan origination than single male borrowers. A higher house value at the time of loan origination appears to increase the reverse mortgage termination probability. Perhaps borrowers with high house values can extract additional housing equity by selling their homes, while borrowers with low house values cannot. In column (2), I controlled for state fixed effects to rid of heterogeneity across states. The results are largely unchanged.

Compared to the probit model, a hazard model takes into account censoring and time-varying explanatory variables. I estimated a Cox proportional hazard model

\displaystyle \lambda_{iat,s}=\lambda^0_s \displaystyle \ast \displaystyle \exp(\beta_1 Term_{iat} + \beta_2 Tenure_{iat} + \gamma_1 Female_{iat} + \gamma_2 Couple_{iat}  
  \displaystyle + \displaystyle \delta_1 log(HV_{iat,s}) + \delta_2 h_{iat,s} + \eta_a + \theta_t)  

where  \lambda_{iat,s} is the termination hazard rate  s periods after loan origination for borrower  i of age  a who originated the loan at time  t.  \lambda^0_s is the non-parametric baseline hazard rate.  HV_{iat,s} refers to the real house value  s periods after loan origination. It is calculated using the real house value at origination and the county-level house price indexes.  h_{iat,s} is the real house price appreciation rate  s periods after loan origination in the county where borrower  i lives.

Column (1) of Table 5 shows the estimation results of the above hazard model. Consistent with the probit model results, borrowers who choose the line-of-credit payment option and single male borrowers tend to terminate reverse mortgage loans sooner than other borrowers. Contemporaneous house values and house price appreciation rates have positive effects on the termination hazard rate. This finding suggests that when house price rises, borrowers are tempted to withdraw additional housing equity by selling their homes. In column (2), I control for state fixed effects and the qualitative results remain the same.

One of the most prominent features of the HECM program is the FHA insurance component. Under this insurance mechanism, borrowers pay an upfront as well as an ongoing mortgage insurance premium. In return, lenders are allowed to assign a HECM loan to HUD when the loan balance reaches 98% of the MCA that was specified at the time of loan origination. If sales proceeds fall short of the outstanding loan balance when the borrower dies or moves away permanently, HUD shoulders the loss - the difference between the loan balance and sales proceeds. Because loans assigned by lenders to HUD often incur a loss for HUD, assignment outcomes are important indictors of the long-term viability of the HECM program.12

Similar to studying the termination outcome before, I examined the assignment outcome using both a probit model and a Cox proportional hazard model. The estimation results are shown in column (3) and (4) of Table 4 and Table 5. Both models suggest that loans with term or tenure payment plans are much less likely to be assigned from lenders to HUD than loans with the line-of-credit payment plan. For example, the hazard model estimates indicate that term loans are 20% less likely to be assigned than line-of-credit loans, and tenure loans are 80% less likely to be assigned than line-of-credit loans. Given that tenure plan monthly payments are calculated assuming the borrower keeps receiving money until her 100th birthday, it is not surprising that tenure loans have such a low assignment hazard. The probit model shows that loans made to single female and married couples tend to be assigned more often than loans made to single males. However, these estimates are no longer statistically significant in the hazard model. In addition, house values and house price appreciation rates appear to have little effect on loan assignment outcomes.

Under the FHA insurance program, borrowers pay 2% of the MCA as the upfront mortgage insurance premium regardless of what payment plan they choose. The estimation results shown above suggest that line-of-credit loans impose significantly higher risks of financial losses on HUD. As a result, there is likely cross-subsidization from reverse mortgage borrowers who choose term or tenure payment plans to those who choose the line-of-credit payment plan. Pricing the insurance premium based on payment plans may reduce such cross-subsidization and decrease the risk of financial losses for HUD.

6 House Price Appreciation

During the first decade of the HECM program, the number of loans originated each year is relatively small and stable. Between 2001 and 2006, the reverse mortgage market experienced substantial growth. The same time period also witnessed an unusual housing market boom across the country. In this section, I investigate the link between local house price movements and reverse mortgage growth.

The amount of loan that a reverse mortgage borrower can access is an increasing function of the appraised house value. When house prices go up, elderly homeowners can borrow more against their homes. For example, Table 1 shows that a homeowner who is 75 years of age with a $200,000 home can pocket $107,715 at the time of loan origination, assuming an expected interest rate of 7 percent. Suppose housing prices in the area rises by 25% and the homeowner's home is now appraised at $250,000. Under the same assumptions about interest rate and upfront transaction costs, he or she can now take home $137,165. The additional $30,000 makes reverse mortgages more attractive. Therefore, all else equal, house price appreciation rates should have a positive effect on the demand for reverse mortgages.

To estimate this effect, I collapsed the loan-level data into county-year level data. I then regressed the percentage change in the number of reverse mortgage loans between time  t and  t+1 in each county on the real house price appreciation rate between time  t-1 and time  t.

\displaystyle \left( \frac{\text{Orig}_{c,t}-\text{Orig}_{c,t-1}}{\text{Orig}_{c,t-1}}\right) = \alpha + \beta h_{c,t-1} + \eta_c + \theta_t + \epsilon_{c,t}    

I controlled for county fixed effects  \eta_c and year fixed effects  \theta_t. To make sure that the dependent variable is measured meaningfully, I limited the sample to counties that have at least 50 reverse mortgages at time  t-1.13

Column (1) of Table 6 displays the estimation results. The estimated coefficient on the real house price appreciation rate is positive and statistically significant, suggesting that places with rising house values tend to have more reverse mortgages originated in the following period. The sample average of the dependent variable is 38.6, meaning that on average the number of reverse mortgages within a county increased by 38.6% per year in the analysis sample. The sample average of  h_{c,t-1} equals 4.0, meaning that, on average, the county-level real house price appreciation rate is 4.0 percent per year. The magnitude of the estimate suggests that a 1 percentage point increase in the annual real house price appreciation rate within the county induces the number of reverse mortgages to grow by 2.5 percentage points. In column (2) and (3), I added lags of the real house price appreciation rate. It appears that the growth rate in reverse mortgages responds most significantly to last year's house price appreciation rate. The housing market movement from three years ago no longer has a noticeable effect on the demand for reverse mortgages.

From 2001 to 2006, median house values in the United States increased by approximately 10% a year in real terms. During the same period of time, the number of reverse mortgages grew by almost 9 times, or 50% a year. The estimates shown in Table 6 imply that a house price appreciation rate of 10% raises the reverse mortgage growth rate by 25%. Therefore, the model estimated above suggests that house price appreciation accounts for about one-half of the overall growth in the reverse mortgage market between 2001 and 2006.

Because a higher house value means that more liquid assets can be extracted from the same physical house through the HECM program, elderly homeowners should find reverse mortgages most desirable when the local housing market is at its peak. To a certain extent, the FHA insurance program insures against the risk of significant house price drops after a HECM loan is originated. If local residents have private information on the future house price movement in the area, they may take advantage of the FHA insurance program and borrow reverse mortgage loans when they expect their house price to decline. To test this hypothesis, I add the house price appreciation rate from time  t to  t+1 to the regression model.14

\displaystyle \left( \frac{\text{Orig}_{c,t}-\text{Orig}_{c,t-1}}{\text{Orig}_{c,t-1}}\right) = \alpha + \beta h_{c,t-1} + \gamma h_{c,t} + \eta_c + \theta_t + \epsilon_{c,t}    

Column (4) of Table 6 displays the estimation results. The estimated coefficient on next year's house price appreciation rate is negative and statistically significant. The magnitude of the estimate suggests that a 10% increase in next year's house prices is correlated with a drop of 7.4 percentage points in reverse mortgage originations. This finding is consistent with the hypothesis that reverse mortgages are most desirable when the local housing market reaches its peak.

In summary, evidence shown in this section suggests that elderly homeowners in areas where the housing market is at its peak are more likely to borrow reverse mortgages. A run-up in previous house prices leads to increases in reverse mortgage originations, which in turn are correlated with declines in house price appreciation rates. A large literature has pointed out that many older households in the United States lack basic financial knowledge and make mistakes in saving and planning for retirement. For example, Lusardi and Mitchell (2005) find that financial illiteracy is widespread among older Americans. The evidence shown in this section seems to suggest that the opposite is true in the reverse mortgage market.

7 FHA Mortgage Limits

As mentioned before, the amount of housing wealth an elderly homeowner can borrow against his or her home is the Maximum Claim Amount (MCA). MCA equals the appraised house value or the county-specific FHA mortgage limit on HECM loans, whichever is smaller. The FHA mortgage limit usually is close to the median house value in the county, but a national ceiling applies to counties with very high median house values and a national floor applies to counties with very low median house values. For example, the national ceiling in 2007 was $362,790. Therefore, all counties with median house values above $362,790 had an FHA mortgage limit of $362,790 in 2007. These limits effectively restrict how much housing equity can be transformed into liquid assets for HECM borrowers. For example, a borrower with a house worth $500,000 could borrow against at most $362,790 of his or her housing wealth. However, a borrower with a $300,000 could potentially borrow against all of his or her housing wealth. When a borrower who wants to tap into her housing wealth chooses between selling the house and taking out a reverse mortgage, both of which have large fixed costs, limitations on how much housing equity she can borrow against may make reverse mortgage a less desirable option to her than selling the house. In this section, I examine whether the FHA mortgage limits reduce the demand for reverse mortgages in areas with very high house prices.

To estimate the effect of the FHA limits, I used a difference-in-differences framework and compared counties where the national ceiling is binding with counties where the national ceiling is not binding. When house prices increase in a county, we expect to see reverse mortgage originations increase accordingly. However, the increase in reverse mortgage origination should be smaller in counties where the national ceiling is binding. For example, in a county where the median home value is well above the national ceiling, homeowners can only borrow up to the national ceiling. Additional housing equity caused by house price appreciation does not translate into more consumable housing wealth for potential reverse mortgage borrowers in this county. More specifically, I estimated the following model

\displaystyle \left( \frac{\text{Orig}_{c,t}-\text{Orig}_{c,t-1}}{\text{Orig}_{c,t-1}}\right) = \alpha \displaystyle + \displaystyle \beta h_{c,t-1} \ast \mathbf{1}(Above Ceiling)_{c,t-1} + \gamma h_{c,t-1}  
  \displaystyle + \displaystyle \delta \mathbf{1}(Above Ceiling)_{c,t-1} + \eta_c + \theta_t + \epsilon_{c,t}  

where  \mathbf{1}(Above Ceiling)_{c,t-1} indicates whether the median house value in county  c at time  t-1 is above the national ceiling for HECM loans at that time. It is a measure of whether the national ceiling is binding in that county. If FHA limits reduce the demand for reverse mortgages in high house value areas, then  \beta would be negative.

Column (1) of Table 7 shows the estimation results. Consistent with the findings in the previous section, real house price appreciation rates in the last period have a large and positive effect on reverse mortgage originations. Counties where the national ceiling is binding have fewer reverse mortgage originations on average, although the effect is not statistically significant. The coefficient of interest,  \beta, is estimated to be negative but very small and statistically insignificant, which suggests that FHA mortgage limits on HECM loans have little effect on reverse mortgage origination. In columns (2) and (3), I tried alternative measures of whether the national ceiling is binding in a county. For example, I used an indicator variable that equals 1 if the county's median home value is at least 20% above the national ceiling in column (2). In column (3), I used the amount of the median home value above the ceiling in percentage terms as a continuous measure. Regardless of which measure I use, the estimated coefficient on the interaction term is small and statistically insignificant. Therefore, I find little supporting evidence that FHA mortgage limits held back the growth of the reverse mortgage market.

Given that only about 1% of eligible elderly homeowners actually took out reverse mortgage loans, perhaps it is not surprising that FHA limits do not seem to have a noticeable effect on the reverse mortgage market. Even in counties where the national ceiling is binding, a great number of elderly homeowners have homes worth less than the national ceiling. The fraction of homeowners with house values below the national ceiling in these counties may be smaller than in other counties, but it is still much larger than the fraction of homeowners who actually purchased reverse mortgages. As reverse mortgages become a more popular financial product among elderly homeowners, the effect of FHA limits may become more important. The American Recovery and Reinvestment Act of 2009 raised the national lending limit for HECM loans to $625,000 only until December 31, 2009. Anecdotal evidence suggests that such an increase in the national ceiling has not induced significant increases in the number of reverse mortgages on expensive homes.

8 Conclusion

The HUD-sponsored HECM program accounts for most reverse mortgages originated in the United States. In the 2000s, the number of HECM loans made each year has been growing substantially. In this paper, I carry out empirical analyses on all HECM loans originated from 1989 through 2007. I find that characteristics of reverse mortgages have changed significantly over time. Borrowers who choose the line-of-credit payment plan, single male borrowers, and borrowers with higher house values exit their homes sooner than other reverse mortgage borrowers. Elderly homeowners are more likely to purchase reverse mortgages when the local housing market is at its peak. Additionally, the FHA mortgage limits do not appear to have impeded the growth of the reverse mortgage market thus far.

Researchers and potential borrowers have had various misconceptions about how the HECM program works. For example, some people think that once a borrower enters the HECM program, the borrower relinquishes his or her house entirely to the lender. In fact, the borrower pays back the lesser of the loan balance or proceeds from the property sale. Hence, the borrower remains the residual claimant on the value of the property, and the moral hazard problem of home maintenance may not be as severe as many people believe.

Another common misconception about the HECM program is that the tenure payment plan, which gives borrowers equal monthly payments for as long as they are alive and continue living in their homes, is equivalent to an annuity. For immediate life annuities, insurance against outliving one's assets is provided by pooling mortality risks across a group of people. However, the tenure plan of HECM loans involves little risk-pooling: If a borrower dies shortly after his or her HECM loan is originated, the borrower pays back only the loan balance, which presumably is small. HUD does not inherit this borrower's entire housing equity to pay another borrower who lives to be over 100 years old. Thus, the longevity insurance aspect of a tenure HECM loan is very limited. In addition, only 10% of HECM borrowers choose the tenure payment plan or the modified tenure payment plan, which suggests that the annuity aspect of reverse mortgages is irrelevant to most borrowers.

Furthermore, many people believe that the high costs associated with HECM loans indicate that these loans are a bad deal for elderly homeowners. One reason that a reverse mortgage costs more than a regular forward mortgage is that HUD charges an MIP to insure both the lender and the borrower. Precisely because there is little risk-pooling in the HECM program, insurance premiums have to be high for HUD to break even. As a result, one may have to change the fundamental structure of the HECM program to cut the costs significantly. It is hoped that the descriptions and analyses presented in this paper will help correct these misconceptions.

References

Bayer, A.-H. and L. Harper. 2000. Fixing to Stay: A National Survey on Housing and Home Modification Issues. AARP Research Report, May 2000.

Bernheim, D.B., L. Forni, J. Gokhale, and L.J. Kotlikoff. 2000. How Much Should Americans Be Saving for Retirement? American Economic Review 90:288-292.

Case, B. and A.B. Schnare. 1994. Preliminary Evaluation of the HECM Reverse Mortgage Program. Journal of the American Real Estate and Urban Economics Association 22:301-346.

Davidoff, T. 2006. Maintenance and the Home Equity of the Elderly. UC Berkeley Working Paper.

Davidoff, T. 2008. Illiquid Housing as Self-Insurance: The Case of Long-Term Care. UC Berkeley Working Paper.

Davidoff, T. and G. Welke. 2007. Selection and Moral Hazard in the Reverse Mortgage Market. UC Berkeley Working Paper.

Lusardi, A. and O.S. Mitchell. 2005. Financial Literacy and Planning: Implications for Retirement Wellbeing. Michigan Retirement Research Center Working Paper 2005-108.

Mayer, C.J. and K.V. Simons. 1994. Reverse Mortgages and the Liquidity of Housing Wealth. Journal of the American Real Estate and Urban Economics Association 22:235-255.

Merrill, S.R., M. Finkel, and N.K. Kutty. 1994. Potential Beneficiaries from Reverse Mortgage Products for Elderly Homeowners: An Analysis of American Housing Survey Data. Journal of the American Real Estate and Urban Economics Association 22:257-299.

Michelangeli, V. 2008. Does it Pay to Get a Reverse Mortgage? Working Paper.

Mitchell, O.S. and J.F. Moore. 1998. Can Americans Afford to Retire? New Evidence on Retirement Saving Adequacy. Journal of Risk and Insurance 65:371-400.

Munnell, A.H., M. Soto, and J.-P. Aubry. 2007. Do People Plan to Tap Their Home Equity in Retirement? Center for Retirement Research Brief Number 7-7.

Rodda, D.T., C. Herbert, and H.-K. Lam. 2000. Evaluation Report of FHA's Home Equity Conversion Mortgage Insurance Demonstration. Abt Associates Inc.

Sinai, T. and N.S. Souleles. 2008. Net Worth and Housing Equity in Retirement. J. Ameriks and O.S. Mitchell, editors. Recalibrating Retirement Spending and Saving. The Oxford University Press: Oxford, UK.

Skinner, J. 2007. Are You Sure You're Saving Enough for Retirement? Journal of Economic Perspectives. 21:59-80.

Szymanoski, E.J. 1994. Risk and the Home Equity Conversion Mortgage. Journal of the American Real Estate and Urban Economics Association. 22:347-366.

Szymanoski, E.J. 2007. Home Equity Conversion Mortgage Terminations: Information to Enhance the Developing Secondary Market. Cityscape: A Journal of Policy Development and Research. 9:5-45.

Venti, S.F. and D.A. Wise. 1989. Aging, Moving and Housing Wealth. D.A. Wise, editor. Economics of Aging. The University of Chicago Press: Chicago, IL.

Venti, S.F. and D.A. Wise. 1990. But They Don't Want to Reduce Housing Equity. D.A. Wise, editor. Issues in the Economics of Aging. The University of Chicago Press: Chicago, IL.

Venti, S.F. and D.A. Wise. 1991. Aging and the Income Value of Housing Wealth. Journal of Public Economics. 44:371-397.

Venti, S.F. and D.A. Wise. 2004. Aging and Housing Equity: Another Look. D.A. Wise, editor. Perspectives on the Economics of Aging. The University of Chicago Press: Chicago, IL.


Figure 1: Calculating the Amount of Payment One Can Receive from a HECM Loan

Figure 1: Calculating the Amount of Payment One Can Receive from a HECM Loan. A flow chart labels four boxes from left to right: House Value, Maximum Claim Amount, Initial Principal Limit, and Net Principal Limit. A box labeled as "FHA 203 (b) Limit" comes in between House Value and Maximum Claim Amount. Two boxes labeled as "Age" and "Interest Rate" enter between Maximum Claim Amount and Initial Principal Limit. Four boxes labeled as "Origination Fee," "Closing Costs," "Initial MIP," and "Servicing Fee" exit between Initial Principal Limit and Net Principal Limit.


Figure 2: Growth in HECM Loans, 1989-2007

Figure 2: Growth in HECM Loans, 1989-2007. This chart plots the number of HECM loans originated by year from 1989 to 2007. Before 2001, the annual origination was below 10,000. From 2001 to 2007, the annual origination increased from around 10,000 to over 100,000.


Figure 3: Fraction of Indebted Homeowners Aged 62 or Above, 1989-2007

Figure 3: Fraction of Indebted Homeowners Aged 62 or Above, 1989-2007. The figure shows three lines. The line labeled by triangles indicates the fraction of elderly homeowners with mortgages, home equity loans, or home equity lines of credit. It increases from 0.22 to 0.39. The line labeled by circles indicates the fraction with credit card debt. It increased from 0.22 to 0.32. The line labeled by squares indicates the fraction with any type of debt. It increased from 0.44 to 0.57.
Note: Data are from 1989, 1992, 1995, 1998, 2001, 2004, and 2007 Survey of Consumer Finances.


Figure 4: Distribution of HECM Borrower Age

Figure 4: Distribution of HECM Borrower Age. The top panel shows the distribution of borrower age for loans originated between 1989 and 1999. The distribution peaks at around age 75. The bottom panel shows the distribution of borrower age for loans originated in 2007. Compared to the top panel, the distribution in the bottom panel shifts to the left and has a spike at age 62.


Figure 5: Distribution of the Initial Principal Limit (IPL) as a Percentage of House Value

Figure 5: Distribution of the Initial Principal Limit (IPL) as a Percentage of House Value. The top panel shows the distribution for loans originated between 1989 and 1999. The distribution peaks when IPL is around 55 percent of house value. The bottom panel shows the distribution for loans originated in 2007. The distribution peaks when IPL is around 65 percent of house value.


Figure 6: Reverse Mortgage Originations as a Percentage of Owner-Occupied Units with Householders Aged 60 or Above, 1989-2007

Figure 6: Reverse Mortgage Origination as a Percentage of Owner-Occupied Units with Householders Aged 60 or Above, 1989-2007.  This is a map of the 48 contiguous states.  California, Colorado, District of Columbia, Nevada, and Utah are shaded in the darkest color. Arizona, Connecticut, Florida, Idaho, Maryland, New Jersey, Oregon, Rhode Island, and Washington are shaded in the second darkest color. Delaware, Illinois, Massachusetts, Maine, Michigan, Minnesota, Montana, New Hampshire, New Mexico, New York, Texas, Virginia, Vermont, and Wyoming are shaded with the second lightest color. The remaining states are shaded with the lightest color.


Figure 7: Termination Rates of HECM Borrowers and
Mortality Rates of the General Population

Figure 7: Termination Rates of HECM Borrowers and Mortality Rates of the General Population. The thick solid line represents male HECM borrowers. The thin solid line represents female HECM borrowers. The thick dashed line represents males in the general population. The thin dashed line represents females in the general population. All four lines increase with age. Males are more likely to terminate loans or die than their female counterparts. The termination rates among HECM borrowers are higher than the mortality rates among the general population.
Note: Termination rates of HECM borrowers are based on 1989-2006 loan-level data. Mortality rates of the general population are based on 2004 mortality tables from the National Center for Health Statistics.


Table 1: HECM Payments to a Hypothetical Borrower
Panel A. Principal Limit Factor
Expected Rate Age = 65 Age = 75 Age = 85
0.05 0.649 0.732 0.819
0.07 0.489 0.609 0.738
0.09 0.336 0.472 0.636

Table 1: HECM Payments to a Hypothetical Borrower
Panel B. Net Principal Limit
Expected Rate Age = 65 Age = 75 Age = 85
0.05 $114,188 $131,492 $150,112
0.07 $83,323 $107,715 $134,344
0.09 $67,200 $94,400 $127,200

Table 1: HECM Payments to a Hypothetical Borrower
Panel C. Tenure Plan Monthly Receipt
Expected Rate Age = 65 Age = 75 Age = 85
0.05 $610 $804 $1,221
0.07 $558 $791 $1,238
0.09 $436 $702 $1,184

Table 1: HECM Payments to a Hypothetical Borrower
Panel D. 10-Year Term Plan Monthly Receipt
Expected Rate Age = 65 Age = 75 Age = 85
0.05 $1,234 $1,421 $1,622
0.07 $983 $1,271 $1,585
0.09 $687 $1,039 $1,467

Note: Assume MCA = $200,000, initial MIP = $4,000, origination fee = $4,000, closing costs = $2,000, monthly servicing fee = $30.


Table 2: Summary Statistics of HECM Loans
Panel A. Compare Early Loans with Recent Loans
  Early Loans
(1989-1999)
Recent Loans
2007
2000 Census
(owners 62+)
Median Age 75.0 72.5 72.0
Single Males 13.9% 19.2% 11.8%
Single Females 56.3% 45.2% 35.4%
Couples 29.8% 35.6% 52.8%
Expected Interest Rate 7.51 5.67  
% Term Plans 16.6% 4.9%  
% Tenure Plans 11.9% 8.3%  
% Line-of-Credit Plans 71.4% 86.9%  

Table 2: Summary Statistics of HECM Loans
Panel B. Median House Values in HECM and SCF
  HECM SCF
1989 $110,337 $100,307
1992 $161,053 $103,428
1995 $144,865 $108,819
1998 $137,352 $114,460
2001 $166,215 $140,463
2004 $208,507 $164,611
2007 $222,000 $175,000

Note: 2000 Census numbers are calculated using the IPUMS data. Median house values are in 2007 dollars.


Table 3: The Distribution of HECM Loan across ZIP Codes
  All Originations (1) All MSA Originations (2) 1989-2002 Originations (3) 2003-2007 Originations (4)
Fraction of High School Graduates 2.595** 3.029** 0.537** 2.492**
Fraction of High School Graduates Standard Errors (0.174) (0.433) (0.103) (0.389)
Fraction of College Graduates 0.201 1.121** 0.579** 0.542
Fraction of College Graduates Standard Errors (0.211) (0.434) (0.104) (0.380)
Log (Median Income) -0.336** -0.042 -0.158** 0.116
Log (Median Income) Standard Errors (0.072) (0.122) (0.033) (0.112)
Log (Median House Value) 1.247** -0.071 0.025 -0.096
Log (Median House Value) Standard Errors (0.041) (0.135) (0.025) (0.126)
Fraction of Blacks 1.919** 2.924** 0.359** 2.565**
Fraction of Blacks Standard Errors (0.087) (0.165) (0.037) (0.151)
Fraction of Hispanics 3.374** 1.861** 0.248** 1.613**
Fraction of Hispanics Standard Errors (0.123) (0.270) (0.061) (0.239)
Fraction of Homeowners Aged 60+ -0.814** -2.025** 0.018 -2.043**
Fraction of Homeowners Aged 60+ Standard Errors (0.166) (0.325) (0.072) (0.291)
MSA Fixed Effects N Y Y Y
Mean of Dependent Variable 1.11 1.66 0.30 1.36
 N 28,808 14,048 14,048 14,048
 R^2 0.157 0.332 0.226 0.333
Note: The dependent variable is the number of HECM loans originated as a percentage of owner-occupied units with householders 60 years or older in the ZIP code. Each observation is a ZIP code in the 2000 Census. Robust standard errors are reported in parentheses. * significant at 0.05 level. ** significant at 0.01 level.


Table 4: Probit Model of Five-Year Loan Termination and Loan Assignment Outcomes
  Termination (1) Termination (2) Assignment (3) Assignment (4)
Term Payout Policy 0.010 0.011 -0.356** -0.352**
Term Payout Policy Standard Errors (0.015) (0.015) (0.054) (0.055)
Tenure Payout Policy -0.136** -0.164** -0.882** -0.890**
Tenure Payout Policy Standard Errors (0.016) (0.016) (0.084) (0.086)
Female Borrower -0.190** -0.185** 0.112* 0.116*
Female Borrower Standard Errors (0.015) (0.015) (0.052) (0.053)
Married Couple Borrower -0.361** -0.358** 0.190** 0.195**
Married Couple Borrower Standard Errors (0.016) (0.016) (0.069) (0.070)
Log (Real House Value at Origination) 0.204** 0.117** -0.032 -0.029
Log (Real House Value at Origination) Standard Errors (0.010) (0.012) (0.033) (0.040)
Age Fixed Effects Y Y Y Y
Origination Year Fixed Effects Y Y Y Y
State Fixed Effects N Y N Y
 N 68,006 68,006 47,223 46,919
Note: The dependent variable in columns (1) and (2) is whether the loan terminates within five years of origination. The dependent variable in column (3) and (4) is whether the loan is assigned by the lender to HUD within five years of origination. The numbers shown are the marginal effects on termination and assignment probabilities. * significant at 0.05 level. ** significant at 0.01 level.


Table 5: Proportional Hazard Model of Loan Termination and Loan Assignment Outcomes
  Termination (1) Termination (2) Assignment (3) Assignment (4)
Term Payout Policy -0.042* -0.034 -0.220** -0.211**
Term Payout Policy Standard Errors (0.021) (0.020) (0.039) (0.039)
Term Payout Policy Hazard Ratios [0.959] [0.966] [0.803] [0.810]
Tenure Payout Policy -0.224** -0.235** -1.544** -1.541**
Tenure Payout Policy Standard Errors (0.014) (0.013) (0.050) (0.050)
Tenure Payout Policy Hazard Ratios [0.799] [0.790] [0.214] [0.214]
Female Borrower -0.207** -0.198** 0.057 0.056
Female Borrower Standard Errors (0.011) (0.011) (0.057) (0.056)
Female Borrower Hazard Ratios [0.813] [0.820] [1.058] [1.058]
Married Couple Borrower -0.426** -0.423** 0.058 0.048
Married Couple Borrower Standard Errors (0.013) (0.012) (0.061) (0.059)
Married Couple Borrower Hazard Ratios [0.653] [0.655] [1.060] [1.049]
Log (Real House Value) 0.249** 0.107** -0.040 -0.063
Log (Real House Value) Standard Errors (0.026) (0.023) (0.034) (0.034)
Log (Real House Value) Hazard Ratios [1.282] [1.113] [0.961] [0.939]
Real House Appreciation Rate 0.018** 0.012** 0.003 -0.000
Real House Appreciation Rate Standard Errors (0.002) (0.002) (0.003) (0.003)
Real House Appreciation Rate Hazard Ratios [1.018] [1.012] [1.003] [1.000]
Age Fixed Effects Y Y Y Y
Origination Year Fixed Effects Y Y Y Y
State Fixed Effects N Y N Y
 N 677,870 677,870 668,878 668,878
Note: Standard errors in parentheses are clustered at the county level. Hazard ratios are reported in brackets. * significant at 0.05 level. ** significant at 0.01 level.


Table 6: House Price Appreciation (HPA) and HECM Loan Origination
Effect of HPA on Loan Origination
(Dep. Var. = % change in loan origination in the county)
  (1) (2) (3) (4)
Real HPA in Year  t-1 2.528** 2.141** 1.940** 2.998**
Real HPA in Year  t-1 Standard Errors (0.287) (0.324) (0.366) (0.355)
Real HPA in Year  t-2   1.082** 1.341**  
Real HPA in Year  t-2 Standard Errors   (0.355) (0.409)  
Real HPA in Year  t-3     -0.517  
Real HPA in Year  t-3 Standard Errors     (0.436)  
Real HPA in Year  t       -0.740*
Real HPA in Year  t Standard Errors       (0.354)
 N 880 880 880 880
 R^2 0.399 0.409 0.410 0.405
Note: A county must have at least 50 reverse mortgage loans in the previous year to be included in the sample. A total of 261 unique counties are included in the sample. A full set of year fixed effects and county fixed effects are included in all columns. Standard errors are clustered at the county level. * significant at 0.05 level. ** significant at 0.01 level.


Table 7: Effect of FHA Mortgage Limits on HECM Loan Originations
  (1) (2) (3)
Real HPA 2.538** 2.589** 2.493**
Real HPA Standard Errors (0.344) (0.317) (0.320)
Above Ceiling -2.559    
Above Ceiling Standard Errors (9.032)    
(Above Ceiling)*HPA -0.060    
(Above Ceiling)*HPA Standard Errors (0.358)    
20% above Ceiling   6.453  
20% above Ceiling Standard Errors   (8.423)  
(20% above Ceiling)*HPA   -0.206  
(20% above Ceiling)*HPA Standard Errors   (0.373)  
Amount above Ceiling     -11.285
Amount above Ceiling Standard Errors     (18.586)
(Amount above Ceiling)*HPA     -0.001
(Amount above Ceiling)*HPA Standard Errors     (0.620)
 N 880 880 880
 R^2 0.397 0.397 0.397
Note: The dependent variable is the percentage change in reverse mortgage origination during the past year in the county. A county must have had at least 50 reverse mortgage loans in the previous year to be included in the sample. A total of 261 unique counties are included in the sample. A full set of year fixed effects and county fixed effects are included in all columns. Standard errors are clustered at the county level. * significant at 0.05 level. ** significant at 0.01 level.


Footnotes

* Federal Reserve Board of Governors, hui.shan@frb.gov. I thank Edward Szymanoski and Douglas McManus for providing me with the data. I thank Michael Mulhall for excellent research assistance. I thank Neil Bhutta, Kevin Moore, Raven Molloy, Shane Sherlund, and Bill Wheaton for suggestions and comments. The findings and conclusions expressed are solely those of the author and do not represent the view of the Federal Reserve System. Return to Text
1. Due to the current financial crisis, the private reverse mortgage market has evaporated so that HECM loans represent nearly 100% of the U.S. reverse mortgage market. Return to Text
2. See Skinner (2007) for a review on this topic. Return to Text
3. See Bayer and Harper (2000). Return to Text
4. Other reverse mortgage products include the Home Keeper program offered by Fannie Mae and jumbo reverse mortgage loans offered by private lenders. Return to Text
5. However, borrowers who have delinquent or defaulted on federal debt may not be eligible for HECM loans. Return to Text
6. For married couples, only the age of the younger borrower is taken into consideration. Return to Text
7. See Szymanoski (1994) for detailed discussions on the assumptions that HUD makes to calculate the principal limit factors. Return to Text
8. The Housing and Economic Recovery Act of 2008 established new limits on the loan origination fee for HECM loans. The limit is the greater of $2,500 or 2 percent of the first $200,000 of the MCA, plus 1 percent of the portion of the MCA that is greater than $200,000. The total amount of loan origination fee may not exceed $6,000. Return to Text
9. Borrowers may change their payment plan throughout the life of the loan at a small cost. Return to Text
10. According to industry specialists, most of the payment plan changes are adding a line-of-credit option to existing term or tenure policies. Because HUD does not keep records on the payment plan history of HECM loans, such assertions cannot be verified. Return to Text
11. A ZIP code is dropped if there are less than 200 residents, the homeownership rate is less than 0.1, the median income or the median house value is zero, or there are no homeowners aged 60 or above in that ZIP code. Return to Text
12. Thanks to the house price run-up between the late 1990s and early 2000s, the number of HECM loans that eventually resulted in losses for HUD is small. Consequently, HUD has earned a profit from the insurance program. It will be interesting to see the effect of the recent housing market downturn on the profitability of the FHA insurance program. Return to Text
13. In robustness checks not shown here, I also tried limiting the sample to counties that have at least 30 or 100 reverse mortgages at time  t-1. The results are essentially the same. Return to Text
14. The inclusion of next period's house price appreciation rates on the right-hand side should not cause endogeneity problems considering that reverse mortgage market is only a tiny part of the housing market and unlikely to drive house prices. Return to Text

This version is optimized for use by screen readers. Descriptions for all mathematical expressions are provided in LaTex format. A printable pdf version is available. Return to Text