The Determinants of Foreclosures for

Page created by Geraldine Long
 
CONTINUE READING
Tucker, Joan S., Howard S. Friedman, Joseph
      E. Schwartz and Michael H. Criqui.
      “Parental Divorce: Effects on Individual
      Behavior and Longevity.” Journal of Per-
      sonality and Social Psychology, 1997, 73
      (2), pp. 381-391.

  The Determinants of Foreclosures for
Single-Family Homes in the United
                States
                                                                               Bryan Duling
I. Introduction                                         increased unemployment due to a recession that is
         In the United States, homeownership cre-       imminently looming. This paper analyzes factors
ates numerous benefits for individuals, families        that cause borrowers to default on their mortgages
and society and is the cornerstone of the “Ameri-       and lose their homes to foreclosure.
can dream.” Low foreclosure rates of residential                National foreclosure proceedings have
mortgages and the stigmatism associated with the        been inflated by significant increases in Califor-
term foreclosure are indicative of the value that       nia, Florida, Nevada, and Arizona. These markets
Americans put on owning their homes. Howev-             are dominated by investor loans, which are loans
er, the rate of mortgages entering the foreclosure      to buyers who do not plan on living in the houses.
process during the 2nd quarter of 2007 was 0.65%.       Nationally, home prices have fallen by 3% caus-
This rate is the highest in history, up seven ba-       ing investors to abandon their mortgages, driving
sis points from the first quarter, twenty-two basis     up foreclosure rates. These markets are also dom-
points from 2006 and showing no signs of a de-          inated by subprime loans: deals offered by lend-
crease (Ackerman, 2007). Below, Figure 1 shows          ers to borrowers with blemished credit histories
the percentage of homes that were in foreclosure at     that have higher rates of interest. During this pe-
the end of the quarter from 1995 to 2007. Figure 1      riod of subprime lending, underwriting standards
illustrates how the current number of foreclosures      were lowered and new affordability products such
is close to surpassing the foreclosure totals seen      as extra-long term, interest only mortgages, and
during the post September 11th recession.               loans with low teaser interest rates that balloon af-
         Although still a small percentage of all       ter a few years (hybrid mortgages) were offered
mortgage originations, defaults and subsequent          (Ackerman, 2007). While more than a third of all
foreclosures are large in absolute numbers and          subprime adjustable rate loans are in the previous
produce crushing losses to lenders and investors,       four Southern states, Ohio and Michigan are two
higher finance costs to consumers, and devastat-        Midwestern states contributing to the foreclosure
ing damage to borrowers and homeowners direct-          problem. These troubles are driven by economic
ly affected. The entire macroeconomy is begin-          problems created by job losses in the manufac-
ning to feel the effects with the real possibility of   turing and the auto industries (Ackerman, 2007).

                                 The Park Place Economist, Volume XVI                                           45
Bryan Duling
     Many lenders, such as Countrywide Financial,                                          literature on the topic of foreclosures and related
     have suffered enormous losses after late payments                                     theory. Section III and IV explain the data and then
     and subsequent defaults mounted. General Mo-                                          lay out an empirical framework, respectively. The
     tors, the world’s largest carmaker, is in danger of                                   model includes traditional determinants of fore-
     having to take a $1 billion charge to cover bad                                       closure, current lending practices, and economic
     mortgage loans of its subsidiary, Residential Cap-                                    conditions. With the modeling framework laid
     ital (“Subprime”, 2007).                                                              out, Section V discusses the results from estima-
             Thanks to problems in the subprime mort-                                      tion of the model. Finally, Section VI presents a
     gage sector, the subsequent decline in housing                                        detailed summary of the findings, possible policy
                                                                                                           implications and avenues for future
                                   Figure 1: % of Loans in Foreclosure at End of Quarter
                       1.60
                                                                                                           research.
                       1.40
                                                                                        II. Review of the Literature and
       % of Loans in

                       1.20
                                                                                        Theoretical Model
          1.00
                                                                                        		      Many studies throughout
          0.80
                                                        % of Loans in Foreclosure       the late 20th century have looked
          0.60                                                                          at the determinants of foreclosure.
          0.40                                                                          Traditionally, these determinants
          0.20                                                                          have included unemployment rates,
          0.00                                                                          change in interest rates, loan-to-
                                                                                        value ratios, and home price appre-
                           96 3

                           97 3

                           98 3

                           99 3

                           00 3

                           01 3

                           02 3

                           03 3

                           04 3

                           05 3

                           06 3

                           07 3
                               1

                               1

                               1

                               1

                               1

                               1

                               1

                               1

                               1

                               1

                               1

                               1

                               1
                             Q

                         19 Q

                         19 Q

                         19 Q

                         20 Q

                         20 Q

                         20 Q

                         20 Q

                         20 Q

                         20 Q

                         20 Q

                         20 Q
                        Q

                             Q

                             Q

                             Q

                             Q

                             Q

                             Q

                             Q

                             Q

                             Q

                             Q

                             Q

                             Q
                       95

                                                                                        ciation. In addition, most studies
                  19

                              19

                                         Time (Q)
                                                                                        have been able to look at the micro
                                                                                        side of the housing issue. Howev-
     construction and sales has had an immediate eco-                er, data sources will force my study to take a more
     nomic impact. A number of real-estate agents,                   national level approach. In addition to looking at
     mortgage brokers, investment bankers, movers,                   more of the national side of the problem, I will
     painters, contractors, landscapers, etc. are expe-              examine a couple of new, untested variables as de-
     riencing lower wages and even losing their jobs.                terminants of foreclosure and use the most recent
     Because home sales and moves stimulate purchas-                 foreclosure data.
     es of appliances, electronics, and furniture, giant                          Two complementary theories, option theo-
     chains like Home Depot and Sears have reported                  ry and the ability-to-pay model, can best explain
     sales down around 5%. In addition, auto sales                   the underlings of the current foreclosure problem.
     were down 12% in July 2007 from the previous                    The basic premise of the option-based model is
     year and consumers reported falling home equity                 that borrowers have the option to default on their
     and rising mortgage payments as a main reason                   mortgage during each payment period (Quercia,
     why (Gross, 2007).                                              2005). Borrowers believe they can benefit from
               With all of this damage being done to the             clearing the value of their mortgage off the books
     economy, an important question needs to be asked:               and gain free rent in their home between the time
     What are the key determinants causing the current               they default and the actual event of foreclosure.
     record number of foreclosures? To answer this,                  The negatives of defaulting and going through
     the study will use the most recent national foreclo-            the foreclosure process for the borrower include
     sure data and attempt to find a model with results              giving up the house, costs of moving, losses of
     that can be interpreted to suggest potential policy             attachable assets, lower credit rating, and any psy-
     implications. Section II of the paper describes                 chological or moral costs of defaulting (Hender-
46                                                        The Park Place Economist, Volume XVI
shott, 1993).                                           Therefore, the borrower is not likely to foreclose
        Option theory stresses that when inter-         until the interest rate jumps after a span of a few
est rates are declining/rising, borrowers have the      years.
strong incentive to default/not default, especially              In addition to option-theory of default,
in regions where house prices are declining/rising      another common reason that individuals default
(Elmer, 2007). If home values fall sufficiently,        on mortgages is due to “trigger events.” This
borrowers will “ruthlessly” exercise their option       view is characterized by the ability-to-pay model
to default (Cutts, 2005). The likelihood of default     which says a mortgagor will default whenever the
increases as the market value of equity declines,       mortgagor’s current income after expenditures
which occurs if either mortgage value increases or      falls below the amount of payment to mortgage
property value decreases (Archer, 2002). Howev-         principal and interest (Barth, 1983). As defined
er, even when in the money, borrowers may delay         by Elmer, trigger events are “unanticipated short-
or refuse the default option although it makes fi-      falls in income such that income is no longer suf-
nancial sense to do so, hoping the option becomes       ficient to meet periodic debt obligations” (2007).
even “deeper in the money” (Quercia, 2005; Pen-         The sudden change then “triggers” default on the
nington-Cross, 2006). These borrowers either            mortgage. Therefore, income and expense-related
believe the negatives of defaulting outweigh the        shocks such as job loss, divorce, or a death in the
costs or believe the volatility in the market could     family may lead to insolvency and mortgage de-
make foreclosure even more attractive in the fu-        fault.
ture.                                                            Higher unemployment rates should be an
        A second theme of option-based models           acceptable proxy for labor market conditions and
is that prepayment and default are “competing”          the chance that the borrower himself faces this
risks. If one option is exercised, the other op-        situation. Danis and Pennington-Cross used un-
tion is terminated. The option to prepay, a call        employment rates lagged one month as a proxy
option, will be exercised when the borrower can         for trigger events and found surprisingly that fore-
get a better deal on an interest rate after financing   closures and unemployment are inversely related
costs. Competing options are especially relevant        (2005). This result differs from theory and natural
now because prepayment penalties are common             intuition. In addition, divorce can be a major fi-
with adjustable rate mortgages. The proportion          nancial stressor especially when alimony or child
of subprime loans with prepayment penalties has         support payments are involved. With the addi-
increased from about 50 percent in 1998 to 80 per-      tion of a second house, living expenses tend to
cent in 2000 (Quercia, 2005). Therefore, this fac-      increase dramatically without any corresponding
tor limits a borrower’s ability to call a mortgage      increase in wages. However, while the divorce
and refinance to a lower rate.                          rate doubled between 1965 and 1976, it has since
        When dealing with the option to default         varied by less than 10% and would not appear to
and go through foreclosure, interest rates have a       be a determinant causing the most recent increase
substantial impact. When interest rates are in-         in foreclosures (Henderschott, 1993).
creasing, the cost of borrowing money increases                  An issue that is related to trigger events is
so foreclosures should increase, ceteris paribus.       the financial risk posture of households. Individu-
However, it is likely that there is a lag between the   als choose their preferred level of savings and in-
time the interest rate increases and the time when      surance to meet those unexpected shocks or “trig-
foreclosures becomes amplified. Adjustable rate         ger” events. The likelihood that a trigger event has
mortgages are likely to come with teaser rates,         a severe negative impact becomes greater as the
below market interest rates, for a short period of      individual’s savings or insurance decreases. For
time and then will adjust to the higher market rate.    example, a lender might foreclose on one home

                                 The Park Place Economist, Volume XVI                                            47
Bryan Duling
     and not another if the foreclosed home’s owner          is a cost advantage to lenders who can then offer
     has less readily available savings. The theory of       lower rates to borrowers who accept a limited pe-
     financial risk posture suggests a possible interac-     riod of payment stability either because they don’t
     tion between savings and unemployment rates             plan to live in the house for long time or because
     and their impact on foreclosures. In the United         they are comfortable bearing the interest rate risk
     States, consumer debt has reached historical highs      (Ambrose Lacour-Little, 2005). Because of the
     while the personal savings rate is at all-time lows     interest rate risk borne by borrowers in the face
     (Elmer, 2007). This recent trend of an extremely        of rising interest rates, one would expect that an
     low savings rate may be having a vast impact on         increase in ARMs relative to fixed rate mortgages
     today’s record foreclosures.                            would increase foreclosures.
              On top of foreclosures caused by trigger               In addition, record gas prices may be hav-
     events, many argue that lenders are part of the cur-    ing a negative impact on individuals’ ability to pay
     rent problem because of the recent development          mortgage payments. In recent years, national gas
     of “predatory” lending tools. Lenders and bor-          price averages have shot skyward due to unrest in
     rowers have the option of using fixed-rate mort-        the Middle East, hurricanes, and increased global
     gages, adjustable-rate mortgages, hybrids of the        demand. This has caused a decrease in discretion-
     two, among other options. Adjustable-rate mort-         ary income. Due to American’s inelasticity with
     gages (ARMs), sometimes referred to as a preda-         respect to gasoline in the short-run, individuals
     tory lending instrument, first appeared after the       may face the choice of fueling up their vehicles
     financial deregulation of 1980 and have quickly         and heating their houses or paying their mortgage
     become a popular mortgage alternative. In 1994          payments on time. Some individuals may opt for
     during a period of rising interest rates, 39% of all    putting off the mortgage payment, hoping and bet-
     mortgage originations were ARMs (Ambrose La-            ting that the bank won’t foreclose on their home.
     Cour-Little, 2005). After the ARM share dropped         This study hypothesizes that as national gas prices
     to around 10% during the late 1990’s, it shot back      increase, the number of foreclosure will also in-
     up into its 1994 range during the start of the new      crease.
     millennium. Below, Figure 2 shows the trend of                  Finally, past research has indicated the
     homebuyers financing through adjustable rate            presence of several other possible determinants.
     mortgages.                                              The most direct measure of equity in a home is
              Consumers are attracted to the lower ini-      the loan-to-value (LTV) ratio, which measures
     tial interest rates of the ARM relative to fixed-       the amount of the loan divided by the value of the
     rate mortgages. For example, ARMs may be                property (Elmer, 2007). Theory would indicate
     especially attractive if they allow homebuyers to       that as LTV decreases, that is as the amount of the
     have lower initial payments in areas where hous-        loan decreases relative to the market value of the
     ing prices have appreciated rapidly (Merry, 2006).      home, foreclosure rates decrease. Homeowner
     Buyers may also be attracted to an ARM if they          equity is also affected by the rate of appreciation
     don’t expect to live in the house for a very long       in house prices. High home appreciation acceler-
     time frame. The buyer may take advantage of             ates the buildup of equity by reducing the current
     the low teaser interest rate, hope that home val-       LTV (Elmer, 2007). However, while the national
     ues increase in the short term, and then sell the       housing price index may be increasing to show
     house and cash out the increase in equity before        positive growth in house value, some borrowers
     the teaser rate is increased after the initial two or   may be still at risk of losing their homes due to
     three-year period.                                      regional housing market price declines (Henders-
              Lenders like the ARM because it puts the       chott, 1993). Overall, as the rate of home appre-
     interest rate risk squarely on the borrower. This       ciation increases, foreclosures should decrease,
48                                    The Park Place Economist, Volume XVI
Bryan Duling
ceteris paribus.                                                        III. Data
        Some supply and demand theory is also                                   The data used in this study are obtained
involved in the current foreclosure crisis. With so                     from a variety of sources that cover the time pe-
many houses available on the market, builders and                       riod starting with the first quarter of 1995 and end-
potential sellers have to slash prices. These lower                     ing with the second quarter of 2007. LoanPer-
aggregate prices are causing problems for borrow-                       formance.com provides services detailing many
ers who have home equity loans or were hoping to                        mortgage statistics and provided the data on fore-
“flip” the home, renovate it quickly and sell for a                     closures used in this study. I will use a quarterly
profit. Nationally, house prices have fallen by 3%                      measure of the percentage of all loans in foreclo-
and the amount of unsold homes has risen to the                         sure at the end of the period for the United States
equivalent of a 10-month supply, so selling one’s                       Mortgage Banker’s Association: National Delin-
home to cover the loss is nearly impossible. It is                      quency Survey default.
easy to see and acknowledge the enormous prob-                                  One key determinant this study looks at is
lems that are rampant in the economy currently                          the effect of the type of mortgage on subsequent
and understand how they could become even                               foreclosures. To measure this, I have gathered
worse in the near future (“The Hammer”, 2007).                          data on the percentage of loans originated dur-
                                                                                    ing a quarter that are adjustable rate
                              Figure 2: % of ARM                                    mortgages. Once a month, as part of
    40%                                                                             the Weekly Primary Mortgage Market
    35%
    30%
                                                                                    Survey, Freddie Mac collects the ARM
                                                                                    share of applications, as reported by
ARM share

    25%
    20%                                                                             lenders who participate in the survey.
    15%
                                                             Figure 2: % of ARM
                                                                                    It is assumed that the rest of the per-
    10%
      5%
                                                                                    centage originated can be considered
      0%                                                                            fixed rate mortgages. I have composed
         Q1 6 Q1 7 Q1 8 Q1 9 Q1 0 Q1 1 Q1 2 Q1 3 Q1 4 Q1 5 Q1 6 Q1 7 Q1             these ARM share figures into quarterly
      95     9    9    9    9    0     0      0    0    0    0      0     0
   19     19   19   19   19   20    20     20   20   20   20     20    20           measures.
                                         Time (Q)
                                                                                            Mortgage interest rates were
                                                                                    also collected from Freddie Mac and
                                                                        include both 30-year fixed rates and 1-year ad-
Most of recent literature on foreclosures focuses
                                                                        justable rate mortgages. Freddie Mac’s Primary
specifically on the subprime mortgage market.
                                                                        Mortgage Market Survey also surveys lenders
The subprime market has rapidly expanded dur-
                                                                        each week on the rates and points for their most
ing the last 20 years. The most typical impairment
                                                                        popular 30-year fixed-rate, 15-year fixed-rate, 5/1
of these borrowers is a poor credit history; others
                                                                        hybrid amortizing adjustable-rate, and 1-year am-
have limited or no documentation on their income
                                                                        ortizing adjustable rate mortgage products. The
or provide no down payment (Danis, 2005). Spe-
                                                                        survey is based on first-lien, prime conventional
cifically, Capozza and Thompson find that sub-
                                                                        conforming mortgages with a loan-to-value of 80
prime loans delinquent for 90 days are more than
                                                                        percent. The interest rate spread and ARM mov-
twice as likely to be foreclosed on as prime loans
                                                                        ing average used in this study were composed
(2006). Recent research on the subprime market
                                                                        from this survey.
and the increased lending to this area has been
                                                                                To measure the volatility of interest rates,
extensive and complex. Due to the complexity of
                                                                        I will use the standard deviation in the London
the issue and the cost of accessing data related to
                                                                        Interbank Offered Rate (LIBOR index). The LI-
subprime lending, this study will not focus on this
                                                                        BOR is the rate of interest at which banks offer to
segment of the problem.
                                     The Park Place Economist, Volume XVI                                                       49
Bryan Duling
     lend money to one another in the wholesale mon-         American housing. Below, Table 1 provides de-
     ey markets in London. It is a standard financial        scriptive statistics on the dependent and explana-
     index used in U.S. capital markets and is the index     tory variables used in the model.
     that is used to set the cost of various variable-rate
     loans. Lenders use such an index to adjust interest     IV. Empirical Model
     rates as economic conditions change. Changes in                 In order to best explain the determinants of
     standard deviation of the LIBOR index will show         foreclosure, I will use the following independent
     any shocks and volatility in interest rates.            variables defined in Table 2: percentage share of
              Other measures of data come from various       mortgage originations that are ARM, interest rate
     sources. The measure of housing prices comes            spread, moving average of 1-year adjustable inter-
     from the Office of Federal Housing Enterprise           est rate, ARM interest rate volatility, house price
     Oversight (OFHEO) that publishes a quarterly            appreciation, national unemployment rate, na-
     House Price Index (HPI). Data on national unem-         tional savings rate, national retail gas prices, and
     ployment rate (Bureau of Labor), national savings       savings and unemployment rates interaction. The
     rate (US Department of Commerce), and average           dependent variable in my study is the quarterly
     retail gas prices (US Department of Energy) were        percentage of homes in the United States in the
     obtained from various government bureaus. All           foreclosure process.
     of the above data was in weekly or monthly form                 I expect to find a positive sign for ARM
     and will be converted into quarterly measures for       share because recent lending practices have creat-
     the purpose of this study.                              ed an environment where lenders offer relatively
              Finally, a measure of loan-to-value will       lower “teaser” rates on adjustable rate mortgag-
     not be included because data on LTV ratios are          es. These teaser rates are priced at below market
     difficult to obtain on the national level and is bet-   interest rates for a period of two to three years.
     ter suited for microanalysis. The use of the quar-      Then the rates reset to the market rate, which is un-
     terly House Price Index should be an acceptable         doubtedly higher, creating more costly mortgage
     proxy for the measure of changes in the equity of       payments. If the environment is such that interest

50                                    The Park Place Economist, Volume XVI
Bryan Duling
rates are rising overall,
when the adjustable
rate becomes fixed af-
ter the period of two
or three years and is
readjusted to a higher
market rate, mortgage
payments will shoot
upward and the like-
lihood of default and
eventually foreclosure
will increase. This po-
tential impact on fore-
closures is most likely
to be seen after a lag of
two or three years. It is
unlikely that many in-
dividuals will want to
default on a mortgage                                    expect the sign on this variable to be positive indi-
priced at a discount, but will be more likely to de-     cating that higher national levels of employment
fault when the interest rate on their loans becomes      are correlated with higher rates of foreclosure.
much higher after the introductory, low teaser rate      This relationship should be seen after a lag in time
period.                                                  because one would expect that individuals have
         To look at the effect of interest rates on      some savings to continue to make mortgage pay-
foreclosure, I will examine three components: the        ments for a period of time after incurring job loss.
interest rate spread, adjustable mortgage interest       In theory, banks would not necessarily foreclose
rate average, and the volatility of ARM rates. In        immediately, but would instead wait and hope the
my model, I will include the interest rate spread        unemployed individual could rejoin the work force
between the prevailing 1-year ARM rate and the           and resume making mortgage payments. For that
30-year fixed rate to proxy for the benefits of          reason, I will lag the unemployment variable one
switching from one type to the other. To measure         quarter.
the effects of interest rates, I will simply use a                I expect the signs on both the national sav-
moving average of the ARM interest rate encom-           ings rate and house price appreciation measure
passing the last four quarters. Moving averages          to be negative. As national savings increases,
are a good tool to smooth out short-term fluc-           people should have more in reserve if a trigger
tuations, thus highlighting longer-term trends or        event does occur or their mortgage payment does
cycles. Using the last 4 quarters should allow a         increase. However, as consumption increases rel-
large enough passage of time to account for read-        ative to saving, the likelihood of a trigger event
justments of ARM into the changing market rates.         or increased mortgage payments having a detri-
Finally, I will test the impact volatility of interest   mental effect increases. Like the unemployment
rates has on foreclosures by using the standard de-      rate, I expect the national savings rate to have a
viation of the LIBOR rate over the period.               lagged effect on foreclosures, so the variable will
         The percent of unemployed workers as            be lagged by one quarter.
provided by the Bureau of Labor is used as a                      Likewise, as the national House Price In-
proxy for the impact of trigger events. I would          dex (HPI) increases, the relative value of Ameri-
                                  The Park Place Economist, Volume XVI                                           51
Bryan Duling
     cans’ homes increase. This increases the equity in       Average, Interest Rate Volatility, House Price Ap-
     homes and lowers the relative value of the mort-         preciation, National Savings Rate, National Retail
     gage and should lower the propensity to default          Gas Price)
     and have one’s mortgage foreclosed. However,
     exceptions can occur with both savings and house         1. As the ratio of ARM to FRM loans increases
     price variable on a regional level as home prices        (ARM share), foreclosures will increase.
     and savings rate do vary based on location and           2. An increase in unemployment rates (lagged)
     preferences. Again, the impact of the HPI may            will increase the number of foreclosures.
     be significant after a lag as it takes a while for ho-   3. An increase in national savings rate (lagged)
     meowners to realize significant changes in home          will decrease the number of foreclosures.
     equity.                                                  4. As housing appreciates nationally, the number
             Finally, the sign on the gas price variable      of foreclosures will decrease.
     is expected to be positive. As national gas prices       5. As national retail gas prices increase, foreclo-
     increase, less money is available for individuals        sures will increase.
     to spend in other areas, including mortgage pay-
     ments. This is due to Americans’ current inelas-         V. Results
     ticity of demand on gasoline. Because so many                    The regression results, presented in Table
     Americans rely on their vehicle for transportation       3, show mixed results in relation to my testable
     to and from work, school, family, events, etc., it is    hypotheses. Table 3 presents the effects of the
     not easy to just lower gasoline consumption when         explanatory variables on the dependent variable
     prices increase. The added cost to buy gasoline          (foreclosures).
     and related products could adversely lead families               Models A and B show the results from
     into foreclosure.                                        my two best regressions. The results show that
             This study will test the hypotheses below        foreclosures are dependent on the lagged share of
     by using the equation:                                   adjustable rate mortgages originated, current un-
                                                              employment, moving average of adjustable mort-
     Foreclosure rate = f(ARM share, Unemployment             gage rates, and the house price appreciation index
     rate, Interest Rate Spread, Interest Rate Moving         lagged one year. Personal savings rate, interest
                                                                                         rate spread, and ARM
                                                                                         interest rate volatility
                                                                                         were not found to be
                                                                                         significant     determi-
                                                                                         nants of foreclosures
                                                                                         by my model. Gas
                                                                                         price was found to be
                                                                                         significant in a few of
                                                                                         the regressions, but its
                                                                                         results were anything
                                                                                         but robust. Therefore,
                                                                                         all of the above vari-
                                                                                         ables were removed
                                                                                         from Models A and B.
                                                                                                 All of the coef-
                                                                                         ficients in Table 3 have
                                                                                         the predicted sign and
52                                     The Park Place Economist, Volume XVI
Bryan Duling
are significant at least to x=0.1. The difference in     study. The Durbin-Watson test is the most com-
lag time of the ARM share is what makes up the           mon way to test for autocorrelation. After running
difference in Model A versus Model B. Model A            the Durbin-Watson test on Model B, it shows that
contains an ARM share lag of two years, while            this model is narrowly within the test limits, and
B uses a lag time of three years. Model A yields         autocorrelation is not present. However, the test
an adjusted R-squared value of 0.680, indicating         showed that autocorrelation is present in Model A.
that 68.0% of the variation in foreclosures is ex-       When the Prais-Winsten test was used to correct
plained by the model. Model B explains 78.0% of          Model A for autocorrelation, both the ARM share
the variation in foreclosures. The variable coef-        and House Price Appreciation variables were no
ficients in Model B are slightly stronger than in        longer significant. Because the only difference
Model A and the adjusted R-squared is signifi-           between Models A and B is the difference in ARM
cantly higher, both of which further the argument        share lag time, Model B with the three-year ARM
that a three-year lag in ARM share origination ex-       share lag continues to look like the best represen-
plains a greater number of foreclosures.                 tative model for the determinants of foreclosure.
         Through the results shown by the coef-                  In relation to my five testable hypotheses,
ficient on the interest rate moving average vari-        I have found mixed results. The results involv-
able, one can see that an increase in the interest       ing ARM share, unemployment, and house price
rate increases the number of foreclosures. At first      appreciation variables came out virtually as pre-
glance, this result appears to be in conflict with the   dicted.
hypothesis generated by option theory that states        1.) An increase in the ARM share originated will
borrowers will want to default when interest rates       cause foreclosures to significantly increase in
are falling not rising. However, option theory as-       three years.
sumes a fixed-rate mortgage, which would give            2.) An increase in the national unemployment
buyers incentive to default when interest rates are      rate will have an immediate, significant impact on
declining, much the same as a company has the            foreclosures.
incentive to call bonds back when interest rates         3.) An increase in the HPI index will cause fore-
fall below the coupon rate. However, my empiri-          closures to decline.
cal model is set up so mortgages are assumed to                  This study predicted that the increase in
be adjustable-rate. Therefore, the borrower does         foreclosures would be seen after a lag in the in-
not hope for rising interest rates because that will     crease in the unemployment rate. However, the
only result in higher mortgage payments.                 increase is seen immediately as the current rate in-
         However, the ability-to-pay model and           creases. This suggests the immediate devastation
more specifically, the impact of trigger events on       a recession could cause on the housing market and
foreclosures were supported by the results of this       the lives of those who do lose their income due to
study. As current unemployment rates increase,           job reductions. Finally, the impact of a change in
more people are unable to meet their mortgage            the HPI index on foreclosures was not as immedi-
payments and subsequently have their homes               ate as predicted, but significant after a one-year
foreclosed. Also, my results agree with the work         lag. This result makes sense because it takes time
done by Elmer who found that when home prices            to realize gains in equity.
are appreciating, less people have their homes                   On the other hand, the two other hypoth-
foreclosed due to increased equity build-up.             eses did not turn out as predicted.
         One potential problem with my data se-          1.) The national, personal savings rate does not
ries is the possibility of autocorrelation occurring     have a significant impact on the number of fore-
in the models. Autocorrelation will often arise          closures.
when time-series data is used, as is the case in this    2.) Retail gas prices do not have a significant im-

                                  The Park Place Economist, Volume XVI                                          53
Bryan Duling
                                                              they have a relatively greater amount in reserves
                                                              such as savings.
                                                                       While the national savings rate alone did
                                                              not turn out to be a significant determinant in hous-
                                                              ing foreclosures in this study, I would not discount
                                                              its possible impact on the economy in the future.
                                                              Unless future savings rates differ greatly from the
                                                              current trend, they could have a significant impact
                                                              on the housing market and entire economy in the
                                                              near future due to the possible interaction with un-
                                                              employment shown in Model C. The same can be
                                                              said for retail gas prices. Again, while not found
                                                              to be significant in this study, current prices al-
                                                              most daily reach all-time inflation adjusted highs.
                                                              If this trend continues, the impact could hamper
     pact on the number of foreclosures.                      the economy, specifically the housing market in
             While my study hypothesized that the im-         the near future.
     pact of unemployment on foreclosures would be
     lagged, the impact of unemployment was signifi-          VI. Conclusions
     cant immediately, suggesting a possible interac-                 The findings of this study extend the previ-
     tion between unemployment and the low current            ous research done on the determinants of foreclo-
     national savings rate. Because the national sav-         sures, specifically with the result of ARM share.
     ings rate is so low, Americans do not have an ad-        The results indicate the significant, negative im-
     equate supply of funds to fall back on when they         pact that adjustable rate mortgages can have on
     lose their jobs. Therefore, I tried to model this        the housing market and economy. Many market
     possibility by creating Model C which contains           analysts have described these adjustable interest
     all the variables previously found significant and       rate lending tools as “predatory.” The results of
     adds in (Savings)(Unemployment), which is an             this study would tend to support that claim. Fu-
     interaction term that multiples the current unem-        ture regulation needs to address the problems with
     ployment rate with the current national personal         adjustable-rate mortgages including the low teaser
     savings rate. Table 4 shows the results when             rates and balloon payments. Adjustable rate mort-
     Model C was run using an OLS regression.                 gages have and will continue to cause problems
             As Table 4 shows, all variables are still        for homeowners unless they are addressed by
     significant at the 0.05 level and all coefficients are   government regulation.
     similar to previous models. Model C explains an                   In addition, this study found that unem-
     impressive 81% of the variation in foreclosures          ployment has a significant, positive effect on fore-
     and is not influenced by autocorrelation. The new        closures. This result goes along with theory and
     (Savings)(Unemployment) interaction variable is          natural intuition but against the results found by
     negative and significant. This suggests that unem-       a previous study done by Danis and Pennington-
     ployment’s positive influence on foreclosures is         Cross. This study also found a potential interac-
     mitigated when unemployment is interacted with           tion effect between savings and unemployment
     an increase in national savings rates. This sup-         rates on foreclosures. The result of this interac-
     ports the theory of financial risk posture that says     tion would indicate that in a country such as the
     a trigger event, namely job loss, will have a much       United States with low savings rates, a recession
     smaller effect on an individual’s financial health if    could have more devastating effects on the econ-
54                                     The Park Place Economist, Volume XVI
Bryan Duling
omy, especially in the area of home foreclosures,       Archer, Wayne R., Elmer, Peter J., Harrison,
than in a country with relatively higher savings              David M., & David C. Ling. “Determi-
rates.                                                        nants of Multifamily Mortgage Default.”
        Future research dealing with the problems             Real Estate Economics, 2002, 30(3), pp.
associated with adjustable-rate mortgages is vi-              445-73.
tal. This study did not even touch on the current
problems in the subprime housing market. Both           Barth, James R., and Anthony M. J. Yezer. “De-
of these areas contain infinite possibilities for re-          fault Risk on Home Mortgages: A Further
search. Possible work comparing and contrasting                Test of Competing Hypotheses.” Jour-
the housing markets of different regions in the                nal of Risk and Insurance, 1983, 50(3),
United States or the determinants of foreclosure               pp.500-5.
between rural and urban areas could bring about
interesting results.                                    Capozza, Dennis. “Subprime Transitions: Lin-
        Finally, the sample size of data used in              gering or Malingering in Default?” The
this study was relatively small. Results could be             Journal of Real Estate Finance and Eco-
enhanced using monthly as opposed to quarterly                nomics, 2006, 33(3), p. 241.
data to increase the number of observations. This
could be fairly easy to accomplish if one has ac-       Cutts, Amy. “On the Economics of Subprime
cess to monthly foreclosure data as that was the               Lending.” The Journal of Real Estate
one variable in this study that was exclusively                Finance and Economics, 2005, 30(2), p.
available in quarterly form.                                   167.
        Foreclosures have become an enormous
problem not just for individual families but also       Danis, Michelle A. and Anthony Pennington-
for the entire well being of the United States econ-           Cross. “The Delinquency of Subprime
omy. Understanding the determinants of foreclo-                Mortgages.” Federal Reserve Bank of St.
sures is vital to protect both individuals and the             Louis, 2005.
economy from the negative consequences this is-
sue triggers. Today, a recession looms due in large     Elmer, Peter J. and Steven A. Seelig. “The
part because of the housing crisis. The emergence             Rising Long-Term Trend of Single-Fam-
of a recession in the United States would exacer-             ily Mortgage Foreclosure Rates.” FDIC
bate problems in the housing market that in turn              Working Paper, 1998.
could initiate a vicious, circular pattern.
                                                        Freddie Mac. “Freddie Mac’s Primary Mortgage
                                                               Market Survey.” 1990-2007. .
     fornia or Subprime?” Forbes.com. 6 Sep-
     tember 2007.                                       Freddie Mac. “Weighted ARM % Average.” Of-
                                                               fice of the Chief Economist, 1995-2007.
Ambrose, Brent W., Michael LaCour-Little,
     and Zsuzsa R. Huszar. “A Note on Hy-               Gross, Daniel. “The New Money Pit.” News-
     brid Mortgages.” Real Estate Economics,                   week, 10 September 2007, p. 28.
     2005, 33(4), pp. 765-82.
                                                        “The Hammer Drops.” The Economist, 4 October
                                                              2007.

                                 The Park Place Economist, Volume XVI                                    55
Bryan Duling

     Hendershott, Patric H., and William R. Schultz.
           “Equity and Nonequity Determinants of
           FHA Single-Family Mortgage Foreclo-
           sures in the 1980s.” American Real Estate
           and Urban Economics Association Jour-
           nal, 1994, 21(4), pp. 405-30.

     “HSH LIBOR.” 1995-2007. .

     Merry, Ellen A., and Michael D. Wilson. “The
           Geography of Mortgage Delinquency.”
           Federal Reserve Board of Governors,
           2006.

     Office of Federal Housing Enterprise Oversight.
            “OFHEO House Price Index.”. 2007. 2Q
            2007 [cited August 30, 2007]. Available
            from www.ofheo.gov.

     Pennington-Cross, Anthony, and Giang Ho.
           “The Termination of Subprime Hybrid and
           Fixed Rate Mortgages.” Federal Reserve
           Bank of St. Louis, Working Papers, 2006.

     Quercia, Roberto G., Michael A. Stegman, and
           Walter R. Davis. “The Impact of Preda-
           tory Loan Terms on Subprime Foreclo-
           sures: The Special Case of Prepayment
           Penalties and Balloon Payments.” Center
           for Community Capitalism, 2005.

     “Subprime Lending: Rising Damp.” The Econo-
            mist, 8 March 2007.

     US Bureau of Labor. “Seasonal Unemployment
          Rate.” 1990-2007.

     US Department of Commerce: Bureau of Eco-
          nomic Analysis. “Personal Savings Rate.”
          1959-2007.

     US Department of Energy. “US Regular Weekly
          Retail.” 1990-2007.
56                                The Park Place Economist, Volume XVI
You can also read