The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
Working Paper Series The Role of Non-Owner-Occupied Homes in the Current Housing and Foreclosure Cycle WP 10-11 Breck L. Robinson University of Delaware and Federal Reserve Bank of Richmond Richard M. Todd Federal Reserve Bank of Minneapolis This paper can be downloaded without charge from: http://www.richmondfed.org/publications/ The Role of Non-Owner-Occupied Homes in the Current Housing and Foreclosure Cycle Breck L. Robinson and Richard M. Todd*† Working Paper No. 10-11 May 2010 Abstract Non-occupant homeowners differ from owner occupants in that they tend to have lower-risk credit characteristics, such as higher credit scores, but may also have weaker incentives to maintain mortgage payments when housing values fall. During the recent housing boom, the share of mortgage borrowing by non-occupant owners was relatively high in states where home values appreciated relatively rapidly. After the housing boom, foreclosures on non-occupant mortgages in several Midwestern and Northeastern states reflected primarily a high rate of foreclosure per mortgage, not a high volume of mortgages to non-occupants. The reverse held true in some coastal and mountain states. Nevada and Florida have experienced the greatest impact overall, because they have both a high volume of mortgages to non-occupant owners and a high rate of foreclosure on those mortgages. Key words: Non-owner occupants, foreclosures JEL Classification: D10, G01 * Breck Robinson is an Associate Professor in the School of Urban Affairs and Social Policy at the University of Delaware and a Visiting Scholar in Banking Supervision and Research at the Federal Reserve Bank of Richmond. Richard Todd is a Vice President at the Federal Reserve Bank of Minneapolis. † For their advice and assistance, we thank Larry Cordell, Anne Davlin, Ron Feldman, Fred Furlong, Michael Grover, Crystal Myslajek, Eli Popuch, Ned Prescott, Dan Rozycki, Michael Schramm, Laura Smith, and Niel Willardson. p. 1 of 69 I. Introduction The past decade’s business cycle was accompanied by a dramatic boom and bust in the housing market. As illustrated in Figure 1, the housing cycle, in turn, was marked by a disproportionate rise and fall in home buying and mortgage borrowing by non-occupant owners, by which we mean individuals who own housing units other than their primary residence. Examples include investors in residential property and owners of vacation homes. As the housing prices declined, foreclosures on properties owned by non-occupants reached high levels in some areas too. In this study, we provide background information on the role that non-occupant homeowners played in the recent housing cycle and foreclosure crisis. We show that the foreclosure rate has been similar for both owner-occupants and non-occupant owners. Nonetheless, non-occupant owners make up a distinct market segment whose activity was associated with regional differences in housing appreciation and foreclosure impacts. We also find that the factors underlying the impact of non-occupant foreclosures vary among states. For example, the high incidence of foreclosures on non-occupant mortgages in Michigan, Indiana and some other states in the Midwest and Northeast primarily reflects a high rate of foreclosure per non-occupant mortgage, not a high volume of mortgages to non-occupants. By contrast, in Hawaii, Idaho, and some other coastal and mountain states, foreclosures on non-owner-occupied homes have been prominent even though the foreclosure rate on non-occupant mortgages has been relatively low. In those states, unlike Michigan or Indiana, the importance of foreclosures on nonowner-occupied properties stems from the high volume of non-owner-occupied mortgages issued. Nevada and Florida stand out because both states experienced a very high volume of mortgages on non-owner-occupied properties and a very high foreclosure rate. II. What Do We Mean by Non-Occupant Owners, and Why Do We Care? Policymakers and journalists have asked whether ―investors‖—those who buy houses not to live in but primarily for expected profits from income and capital gains—have played a disproportionate role in the current housing cycle. However, because it is often difficult to measure investor activity in the housing and mortgage market, we focus instead on the broader concept of non-occupant owners of single-family homes. A single family home is defined as a detached structure with one to four residents or a townhouse and a condominium contained in a larger building but available for sale separately. For the purposes of this study, we will ignore investments in multifamily rental housing, defined here as structures with five or more housing units intended for renters.1 Decisions about how to analyze non-occupant owners are often driven by data limitations. For example, data collected under the Home Mortgage Disclosure Act (HMDA) distinguish between mortgages on owner-occupied and non-owner-occupied properties but provide no additional information as to whether a non-owner-occupied property is a vacation home, an investment property, or something else. However, data obtained from Lender Processing Services (LPS) allow the non-owner-occupied category to be segmented into two groups; second homes and other (or 1 Multifamily housing played a relatively small role in the recent housing cycle. Between 2000 and 2005, the number of multifamily rental housing starts increased by 4 percent, far less than the 38 percent rise in starts of one- to four-unit homes. Between 2005 and 2008, multifamily starts fell by 15 percent, compared to the 64 percent drop in starts of oneto four-unit homes. p. 2 of 69 ―investor‖ properties).2 Second home is defined as a non-primary residence purchased mostly for recreational use or as an occasional or seasonal residence. Although the accuracy of this additional information is unclear, we make some use of it to shed additional light on the role of non-occupant owners in the recent cycle.3 Why should we care about the role non-occupant homeowners played in the current mortgage crisis? One reason is to have a better understanding of housing cycles in general and the recent foreclosure crisis in particular. Given the still very high rate of mortgage defaults in 2010, a better understanding of the role of non-owner-occupied homeowners might also help direct policymakers in their efforts to reduce foreclosures. Current foreclosure mitigation efforts have been used to address the needs of owner-occupiers rather than other owners, but cities struggling to contain the negative spillover effects of vacant properties might benefit from a better understanding of the local prevalence and implications of non-owner-occupant foreclosures. Finally, efforts to improve the future supervision and regulation of housing and mortgage markets will require an accurate assessment of what led to the current crisis. Although much of the current interest in non-owner-occupied homes focuses on the problems that profit-oriented investors may have caused in bidding up home prices or triggering foreclosures, it is worth noting that many non-occupant owners play a beneficial role in the single-family housing market. According to a Harvard University Joint Center for Housing Studies analysis of the 2003 American Housing Survey, 35 percent of American renters live in single-unit properties and another 21 percent live in two- to four-unit structures.4 That is, non-occupant owners of single-family housing provided about half of the rental units in the U.S. in 2003. III. Non-Occupant Owners Are a Distinct Market Segment Non-occupant owners are likely to have different motives, characteristics, and behaviors than owner-occupants. Below, we discuss these differences conceptually and validate the differences in characteristics and behaviors using data from mortgage and foreclosure records. Motives. Homeowners derive two distinct benefits from owning a home, the value associated with consumption and the potential financial benefits that we will call profit. Homeowners derive consumption value by occupying the home and enjoying the direct benefits from its use. Homeowners can also gain financially if their home appreciates or if they rent out all or a portion of the home. Depending on how the homeowner plans to use the home will determine how much value and the type of value that homeowner will derive from the home. At one extreme, owneroccupants would put a relatively high value on the consumption value of home ownership since 2 We use LPS’s mortgage data for much of the analysis in this paper, partly because they provide the only information we have on mortgage performance. In Appendix 2, we consider how the LPS data compare to alternative national and local data sources on mortgages. The results indicate that the LPS data cover only about 60 percent as many mortgages as the HMDA data include and may undercount the ratio of mortgages to non-occupants slightly when compared with the HMDA data. Nonetheless, we also find that the LPS data provide fairly reliable measures of useful ratios and relationships, such as foreclosure rates and rankings of regional outcomes. 3 In Appendix 3, we use the LPS data to examine the differences between residential mortgages to second-home and investor borrowers. In most states, investor borrowers outnumber second-home borrowers by three-to-one or more, but second-home mortgages become more significant in a few high-amenity states like Hawaii. Both HMDA and LPS depend on homeowners to self-report the occupancy status of the home, and LPS also relies on self-reporting to distinguish second homes from investment properties. Because of differences in underwriting based on the occupancy and usage of the home, borrowers have an incentive to misreport their true intentions. 4 We calculated these percentages from the more detailed data presented in Figure 8-1 of Apgar and Narasimhan (2008). Their figures also include information on single-unit homes including manufactured housing units. p. 3 of 69 they are able to use and modify the home to meet their needs. At the other extreme are owners who do not reside in the home and therefore consume very little of the non-monetary benefits of homeownership. These owners place a higher value of the current and future cash flows that may be generated from the home in the form of rental income and capital gains, net of costs. Owners of vacation homes or rental properties who occupy them only occasionally would fall somewhere in between these two extremes. Characteristics and Risk Factors. It makes sense that non-occupant owners would have different financial characteristics than owner-occupants, given the capital and credit needed to purchase an investment or vacation property.5 As shown in Figures 2, 3, and 4, data from mortgage applications validate this supposition. Using HMDA data, Figure 2 shows that non-occupants tend to have noticeably higher incomes than owner-occupants. In the typical state, the median income of nonoccupant home purchasers rose from just over $100,000 in 2004 to close to $125,000 in 2007. By contrast, in the typical state the mean income of owner-occupant borrowers was about $60,000 to $65,000. The incomes of borrowers who refinanced exhibited similar patterns by occupancy status. Using LPS data, figures 3a to 3c show that, by common underwriting standards, non-occupant borrowers have a lower risk profile than owner-occupants. Figure 3a shows that non-occupant homeowners have higher credit scores when compared to owner-occupants. This is especially true of vacation-home buyers, whose average FICO score in a typical state is about 740, compared with about 720–730 for investors and 700 for owner-occupants. Another measure of credit risk is the loan-to-value (LTV) ratio.6 Figure 3b shows that LTV ratios on investor-owned mortgages tended to be comparable to or somewhat lower than LTV ratios on loans to owner-occupants. The lowest LTV ratios were on mortgages on vacation homes. Overall, then, non-occupant mortgage holders have less risky LTV ratios. Mortgage applicants’ total debt relative to their income is another common risk measure. Figure 3c shows that the debt-to-income (DTI) ratio for non-occupants tends to be on par with or lower than the DTI ratio for owner-occupants. Compared with owner-occupants, investors had lower DTI ratios on mortgages originated in 2004-2005 and comparable ratios in 2006-2007. DTI ratios on second-home mortgages were lower in all four years shown. Overall, compared with owneroccupants over 2004-2007, non-occupants tended to have higher credit scores and comparable or lower LTV and DTI ratios. Although the higher incomes and superior credit scores of non-occupant borrowers could give them access to larger amounts of mortgage credit, Figure 4a shows that non-occupant mortgages are slightly smaller in size. When combined with non-occupants’ substantially higher incomes, these lower loan amounts imply significantly lower loan-to-income ratios for non-occupants. The HMDA and LPS data confirm that non-occupants do indeed have lower loan-to-income ratios, another 5 We also looked at the racial, ethnic, and gender differences between owner-occupants and non-occupant owners, using HMDA data for the primary loan applicant. In most states the primary applicant on non-occupant mortgages was more likely to be male. Racial and ethnic differences between occupant and non-occupant primary applicants were generally small with the largest difference between Latino owner-occupants and non-owner occupants in states like Arizona, California, Florida, New Mexico, Nevada, and Texas. 6 The data for loan-to-value is derived at the time the loan is originated. However, the LTV figure maybe different from the original amount if the loan amount or appraisal is updated while the loan is being serviced. As a result, it is possible that some unknown percentage of the mortgages in the sample may have appraisal values that are provided after the loan was originated. As a result, if the non-occupant property values have fallen more (less) rapidly than the value of owner-occupant properties, then data would exaggerate (underestimate) the relative ex ante risk investors. p. 4 of 69 indicator of lower credit risk.7 In short, standard criteria used to underwrite mortgages indicate that non-occupants should be better credit risks than owner-occupants. Behavior. Despite their apparently superior risk credentials, non-occupant owners are often no more likely than owner-occupants to make their mortgage payments. In deciding whether to pay, both non-occupant owners and owner-occupants consider the foreclosure option, whereby their mortgage debt is extinguished and the property’s title is conveyed to the lender. However, nonoccupant owners may be freer to take advantage of the profit opportunities this option creates. Both know that the foreclosure option limits their direct financial loss to the equity they have invested in the home, primarily their down payment.8 Because the homeowner has limited liability, the value of the foreclosure option rises as the owner’s equity (the difference between the value of the home and the amount owed on the mortgage) declines or turns negative. The relationship between the foreclosure option and equity is particularly strong for non-occupants, because non-occupants place a particularly high value on the profit opportunities that may occur associated with homeownership. In addition, foreclosure imposes greater noninvestment losses on owner-occupants, such as moving costs, credit impairment effects, and emotional stress. As a result, owner-occupants are less likely to exploit the mortgage foreclosure option. Not facing these same constraints, non-occupants are freer to exercise their foreclosure option when expected profits are low.9 For this reason, nonoccupant homeowners are considered ―ruthless‖ borrowers.10 The theory behind the mortgage foreclosure option suggests that, other things being equal, foreclosure rates will be higher for mortgages on non-owner-occupied homes. For this reason, mortgage lenders often compensate for the ruthlessness of non-occupant borrowers by imposing stricter terms on their mortgages. A Federal Reserve staff member experienced in banking supervision and policy analysis states that most mortgage lenders charge an interest rate premium on mortgages to non-occupants ranging from 25 to 100 basis points, other things being equal. Lenders may also seek additional risk controls, such as higher down payments on loans to nonoccupants.11 The higher default risk associated with non-occupants is also known among investors in mortgage products. These investors often place restrictions on the percentage of loans originated to nonoccupants in their mortgage portfolios. According to Doug Duncan, the chief economist for the Mortgage Bankers Association, investors expect that ―people who don't live in homes are more 7 However, we cannot tell from the HMDA and LPS data if non-occupants support multiple mortgages with the same reported income. 8 Many states allow creditors to seek additional compensation from the borrower when sale of the property fails to cover the unpaid mortgage balance, but in practice these deficiency judgments are infrequently pursued and even less frequently collected. 9 See Gerardi, Shapiro, and Willen (2008). Non-occupants also have some disadvantages. As noted above, public assistance to delinquent borrowers usually targets owner-occupants and seeks to exclude non-occupants. In addition, whereas owner-occupants in effect rent to themselves, non-occupants must either forgo rental income or seek tenants. The latter involves both administrative costs and risks, such as the tenant damaging the property or failing to pay the rent. When expected capital gains are small, these disadvantages are relatively large and can significantly limit nonoccupant demand for single-family homes. 10 In Appendix 1, these basic motives are presented more formally in a simple adaptation of the mortgage payment model of Gerardi, Shapiro, and Willen (2008). 11 For example, a 1998 filing with the Securities and Exchange Commission (SEC) by Green Point Financial Corporation notes that first-lien adjustable rate mortgages to non-occupants require at least 25 percent down, an elevated interest rate, and possibly a prepayment penalty. See sec.edgar-online.com/1998/03/26/09/0001047469-98011577/Section2.asp. p. 5 of 69 likely to walk away when mortgage rates reset higher or the property loses value.‖12 His assessment is corroborated by filings with the Securities and Exchange Commission that caution investors in mortgage backed securities about the presence of mortgages to non-occupants.13 Previous studies find that default and loss rates are higher on mortgages to non-occupant owners than on mortgages to owner-occupants, at least after controlling for credit scores and other standard underwriting criteria. Cowan and Cowan (2004) use 1995-2001 data provided by a large subprime mortgage lender, and they state that ―The likelihood of default, whether measured by foreclosure or delinquency, is much greater for properties where the owner is not the occupant.‖14 Immergluck and Smith (2004) find that, in the Chicago area, a higher volume of subprime lending to nonoccupant owners is associated with higher foreclosure rates, even when several risk factors are accounted for.15 Avery, Canner, and Brevoort (2007) find that, other factors held constant, nonoccupants’ share of a county’s mortgages in 2004 was significantly associated with the county’s increase in mortgage delinquencies between 2004 and 2007.16 In a detailed analysis of public records in Massachusetts, Gerardi, Shapiro, and Willen (2008) use mortgage data on condominiums and multi-family dwellings to proxy for loans to non-occupants.17 After controlling for other risk factors, they state that ―owners of condominiums and multi-family homes are estimated, respectively, to have 42 percent and 57 percent higher conditional default probabilities than owners of single-family homes.‖ Using data from over 100,000 subprime and Alt-A mortgages that were originated between 1992 and 2007,18 Bajari, Chu, and Park (2008) find that, other things being equal, non-occupants are more likely to exercise their foreclosure option. Haughwout, Peach, and Tracy (2008) compare owner-occupants, investors, and second-home buyers with regard to rates of early (within the first year) default on mortgages originated from 2001 to 2007. After controlling for observed risk factors, these authors find that ―investors are more likely to default in the first year‖ and that investors’ early default behavior is more sensitive to housing depreciation and having negative equity in the property. They also find that owners of second homes behave differently from investors and are not more likely than owner-occupants to default early.19 Lenders may also face higher rates of loss given foreclosure when the property is non-owner-occupied.20 We have seen that non-occupant owners tend to have stronger financial characteristics than owneroccupants but less incentive to make mortgage payments when times are tough. Nationally, these opposing effects have apparently offset each other in recent years, resulting in only small differences in foreclosure rates on mortgages to non-occupant owners and owner-occupants. Our LPS data on home-purchase and refinance mortgages originated in 2004 show that the national foreclosure rate on non-occupant mortgages was 4.5 percent, which is slightly under the 5.0 percent 12 Investors own about one-fifth of Bay Area homes in foreclosure. See McCormick and Said (2007). For example, the S-3/A SEC Filing by J. P. Morgan Acceptance Corp I on February 8, 2006, states ―The mortgaged properties in the trust fund may not be owner occupied. Rates of delinquencies, foreclosures and losses on mortgage loans secured by non-owner occupied properties may be higher than mortgage loans secured by a primary residence.‖ See http://sec.edgar-online.com/2006/02/08/0001162318-06-000172/Section14.asp 14 See Cowan and Cowan (2004). The reported quote is especially unfavorable to investors because it appears to be unconditional; that is, they do not include controls for risk factors such as the borrower’s income or credit score. The authors report a smaller difference in performance between mortgages on vacation homes and owner-occupied homes, in part because in their data (as in our LPS data) owners of second homes tend to be above average in credit quality. 15 See Immergluck and Smith (2004). 16 See Avery, Brevoort, and Canner (July 2008). 17 See Gerardi, Shapiro, and Willen (2008). 18 Personal communication from Sean Chu of the Board of Governors of the Federal Reserve System, based on supplemental calculations related to research reported in Bajari, Chu, and Park (2008). 19 See Haughwout, Peach, and Tracy (2008). 20 See Qi and Yang (2007). p. 6 of 69 13 rate on owner-occupants’ mortgages.21 As defaults escalated for both groups in subsequent years, the foreclosure rate for non-occupants grew somewhat higher but remained close to the foreclosure rate for owner-occupants. By year of mortgage origination, the foreclosure rates for non-occupants and owner-occupants were the following: 8.5 and 8.1 in 2005, 12.6 and 12.0 percent in 2006, and 7.6 and 6.8 percent in 2007. An examination of state foreclosure rates confirms the general alignment of non-occupant and owner-occupant foreclosure rates but also identifies places where they diverge. Figure 5 displays the relative foreclosure rates on non-owner-occupied and owner-occupied mortgages by year of origination for 2004-2007 by state. The 45-degree line in the figure represents equality between the foreclosure rate for non-occupants and occupants. The symbol for most states lies near this line, in keeping with the similarity on non-occupant and owner-occupant foreclosure rates nationally. There are exceptions, however. The symbols for some of the Midwestern states (for example, Michigan, Indiana and Ohio) are well above the 45-degree line, which means that in these states non-occupant borrowers are distinctly more likely than owner-occupants to be in foreclosure. The symbols for some of the rural or Sunbelt states that may have a high percentage of vacation homes relative to the national average are well below the 45-degree line, at least in some years. This means that nonoccupant owners are less likely to enter into foreclosure than owner-occupants. Despite these exceptions, Figure 5 indicates that foreclosure rates between non-occupants and owner-occupants are highly correlated across the 50 states and the District of Columbia.22 We noted above that non-occupant owners may focus more exclusively on the expected profits of home ownership, when compared to owner-occupants. For this reason alone, non-occupants are likely to respond more quickly to changes in economic conditions or demographic trends that would generate profitable opportunities in acquiring single-family housing. In addition, investors and even vacation home buyers can own multiple properties in multiple areas, which is impossible (by definition) for owner-occupants. For these reasons, we expect non-occupants’ buying and mortgage borrowing activity to be more procyclical than for owner-occupants. Using data from the National Association of Realtors (NAR), Figure 6 supports the relative procyclicality of non-occupant home buying activity. As the housing boom rose to its peak in 2005, home buying by non-occupants grew faster than that of owner-occupants. For example, purchases by investors increased almost 50 percent between 2003 and 2005. Over the same time period, investors’ share of the overall increase in home sales is about 61 percent. If you add in the 20 percent rise in vacation-home purchases, non-occupants account for 75 percent of the growth in sales in the final two years (2004–2005) of the boom. In the subsequent housing downturn, the NAR data show that non-occupants accounted for 65 percent of the decline in home sales. Over the same period, the cyclical swing in owner-occupant purchases is much milder. For example, sales to owner-occupants rose just 6.4 percent from 2003 to 2005 and then fell 13.5 percent during the housing downturn. The HMDA data also show that non-occupant owners were relatively active in the recent housing cycle. We focus here on originations of home-purchase mortgages, rather than mortgages used for home improvements or refinancing of an existing mortgage. As reflected in Figures 1 and 7, nonoccupants’ share of home-purchase mortgages had been rising since the early 1990s but accelerated after 2001. As the housing boom peaked, home-purchase mortgage originations to non-occupant 21 Our data include all foreclosures from the date of origination through July 2009. Cross-sectionally, for the 50 states and District of Columbia, the correlations between owner-occupied and nonoccupant foreclosure rates are 0.87 in 2004, 0.90 in 2005, 0.89 in 2006, and 0.92 in 2007. p. 7 of 69 22 owners rose 84 percent between 2003 and 2005. Over the same period, home-purchase mortgage originations to owner-occupants rose much less, 36 percent. As a result, non-occupant borrowers accounted for 26 percent of the increase in home-purchase mortgage borrowing between 2003 and 2005, even though their share of this market was only about half that size.23 Both the NAR and HMDA data report flows, for either new purchases or new mortgages. Neither directly addresses what happened to the stock of housing units. For example, a large flow of nonoccupants buying homes could simply reflect the churning of properties among non-occupant owners, with no net increase in the number of non-owner-occupied properties. A rise in the number of non-occupants who buy properties from developers in order to ultimately resell them to owneroccupants (sometimes referred to as flipping) could also cause a boost in non-occupant buying with little net change in non-occupants’ share of ownership.24 In fact, data from the American Housing Survey (AHS) show that non-occupants’ share of ownership did not parallel their share of home purchasing and borrowing. Although the flow data show a big rise in the share of buying and borrowing by non-occupants in 2003-2005, the AHS data show the stock of owner-occupied single-family homes growing faster than other types of housing units.25 Some of this divergence could be explained by considerable churning or flipping in the stock of non-owner-occupied housing for the period 2003-2005.26 After 2005, the data on single-family housing stock are difficult to interpret, due to a big increase in vacant properties that may reflect foreclosures. Again, the stock data appear to be at odds with the flow data for purchases and mortgages. For example, the AHS data indicate that the stock of nonowner-occupied single-family housing grew faster than the stock of owner-occupied housing in the 23 The NAR and HMDA data disagree over the magnitude of non-owner occupants’ role in the housing boom and bust, although both sources confirm that home purchasing was more procyclical for non-occupant owners than for owneroccupants during the recent housing cycle. We have not looked into the reasons for the differences between the NAR figures and the HMDA data, but there are some obvious possibilities. NAR’s figures come from a survey, which is subject to sampling error and, possibly, response bias. However, HMDA data have limitations as well. HMDA only captures home purchases that involve a new mortgage (thereby omitting cash purchases). In addition, HMDA’s coverage under-represents mortgages by small lenders in rural areas, which could lead to an undercount of vacationhome mortgages, for example. The HMDA data can also exaggerate changes in home purchases, when two or more mortgages are associated with a single home purchase. For example, combinations of first and second mortgages to finance a single home purchase increased substantially in 2004–2006, according to Avery, Brevoort, and Canner (2007). 24 The term ―flipping‖ generally has a negative connotation but can also provide needed financing for housing development in rapidly growing markets. 25 From the second quarter of 2003 to second quarter of 2005, the total number of owner-occupied single-family housing units grew by 3.8 percent, faster than the 2.9 percent increase in total single-family housing units. Thus, owner-occupied single-family units grew faster than the total of all other kinds of single-family units. (We defined a single-family unit as a unit in a structure with 4 or fewer total units.) A 2.2 percent drop in the number of single-family rental units that were occupied or for rent (as opposed to for sale) restrained the overall growth in single-family units between 2003 and 2005. However, rapid growth in certain sub-categories of the non-owner-occupied single-family housing stock from 2003 to 2005 was more aligned with the data on non-occupant buying and borrowing. For example, the stock of seasonal (vacation) single-family properties grew 14 percent over this period, and the stock of vacant single-family properties for sale rose 16 percent. 26 See Benjamin, Chinloy, Hardin III, and Wu (2008) for a model that incorporates residential real estate churning and flipping, with an application to the Miami, Florida, real estate market for the time period 2004-2006 . They argue that apartment buildings were converted to condominiums because investors seeking to convert them were able to outbid investors who were planning to maintain apartments as rental units. They find (pp. 631-632) that there is ―a higher price when the volume of sales accounted for by converters rises‖ and that the ―entry of condo converters to the MiamiFort Lauderdale apartment market in 2004 causes a bubble in prices, which eventually dissipates.‖ p. 8 of 69 years 2005-2007. However, during this same time period, non-occupants accounted for a disproportionate share of the drop in the flow of home purchases and home-purchase borrowing.27 Although the NAR and HMDA flow data show different patterns of change than the AHS housing stock data, they mostly agree on a key point—that the behavior of non-occupant buyers and borrowers is in many ways distinct from that of owner-occupants. In this section, we have presented the case that investors and other non-occupant home owners have motives, characteristics, and behavior patterns that, on average, differ from those of owner-occupants. Having established that they are different, we now look across space (regions of the U.S.) and time to examine how home buying and mortgage borrowing by investors and other non-occupant borrowers were related to housing prices and foreclosures in recent years. IV. Non-Occupant Owners and Housing Prices We argue above that non-occupant home buyers respond primarily to profit opportunities and can do so more quickly and extensively than owner-occupants. This suggests that non-occupant owners may be attracted to areas where single-family housing is expected to appreciate more rapidly. In other words, causality might run from expectations of rapidly rising prices to above-average buying activity by non-occupants. Causality may flow in the other direction too, such as when a growing population of affluent middle-aged families seeks to buy vacation properties, boosting demand and increasing prices for homes in recreational areas. The direction of causality in these relationships is often difficult to determine, and we do not attempt to do so here. However, we can document growing correlation between housing appreciation and non-occupant mortgage borrowing near the end of the housing boom. We wish to observe if non-occupant borrowers are relatively active in areas where housing prices appreciate rapidly. To do so, we spatially correlate the level or percentage change in the share of home-purchase mortgage originations to non-occupants with the percent change in housing prices, across the 50 states and the District of Columbia.28 In Figure 8, we take the percentage change for a ten-year period, for the share of home-purchase mortgage borrowing by non-occupants and housing prices.29 The figure shows that a number of states lie fairly close to the 45-degree line, which would symbolize a one-to-one relationship between the percentage change in non-occupant share and the percentage change in housing prices. However, several states lie far from the 45-degree line. Among these outliers, the more common case is for non-occupants’ share of home purchase mortgages to grow considerably faster than home prices, as in Delaware, Georgia, Idaho, Indiana, Michigan, Minnesota, Montana, New Mexico, Pennsylvania, Tennessee, or Texas. The clearest case of prices rising faster than non-occupants’ share is California, but the District of Columbia, Florida, Massachusetts, and Rhode Island also share this pattern. With so many points lying well off the 45-degree line in multiple directions, the overall statistical correlation is low. 27 From the second quarter of 2005 to second quarter of 2007, the total number of owner-occupied single-family housing units grew by 2.8 percent, which is slower than the 3.4 percent increase in total single-family housing units. Thus non-occupants’ share in the ownership of single-family housing units increased during the housing bust. A 5.9 percent rise in the number of single-family rental units that were occupied or for rent contributed to the overall growth in single-family units between 2003 and 2005 as well as to the rise in non-occupants’ share of ownership. The stock of seasonal (vacation) single-family properties and vacant single-family properties for sale also continued to grow rapidly over this period (by 14 and 40 percent, respectively). 28 The mortgage data come from HMDA. Repeat-sales data for housing from the Federal Housing Finance Authority (FHFA, formerly the Office of Federal Housing Enterprise Oversight) are used to measure housing appreciation. 29 The correlation was just 0.09, which is not significantly different from zero. p. 9 of 69 Although home prices and the share of non-occupant home-purchase mortgage borrowing had only a slight tendency to move together over the 1996–2006 time period as a whole, their relationship tended to be stronger near the end of the period. This is illustrated in Figure 9, which uses the state data mentioned above to compute cross-sectional correlations between the one-year percentage change in home-prices and the share of mortgage originations for home-purchase to non-occupant borrowers.30 The left-most (green) bar for each year displays the contemporaneous (no lags) correlation between the annual percentage change in home prices and non-occupants’ share of home-purchase mortgages. The middle bar (yellow) displays correlations in which non-occupants’ share of mortgages is lagged one year, and the right-most (maroon) bar displays correlations in which the price change is lagged one year. Values above 0.25 or below -0.25 are statistically significant at a 10 percent level or better. Using 10 percent as the standard, the contemporaneous correlations are insignificant or negative until 2003, but then the correlations become mostly statistically significant and positive for the final three years of the boom.31 As a result, we find that as the housing boom progressed, the spatial correlation of home price appreciation and nonoccupants’ share of home purchase mortgage borrowing seemed to increase from weak or negative to moderately positive and statistically significant.32 V. The Spatial Pattern of Non-Occupant Mortgage Borrowing and Foreclosures As noted earlier, foreclosure rates for non-occupant owners and owner-occupants have been similar, both nationally and at the state level, during the housing bust. However, significant variation in the extent of non-occupant borrowing and non-occupant foreclosure rates has made the impact of foreclosures on properties owned by non-occupants far from uniform across the United States. In this section, we document and analyze these regional variations.33 One way to measure the extent of non-occupant foreclosures is to compute the percentage of foreclosures that involve properties owned by non-occupants. Figure 10 uses LPS data for mortgages originated in 2006 to show that this percentage varies significantly across states. For example, the share of foreclosures involving non-occupant owners averages 11 percent nationwide but equals or exceeds 14 percent in two Midwestern states, four Southeastern states, and three Western states. The share of foreclosures involving non-owner-occupied properties may be useful for some purposes, such as determining which types of foreclosure mitigation programs are needed in a given area. Nonetheless, we do not emphasize it in this study because it may distort the relative significance of non-owner-occupant foreclosures across regions. For example, in states with relatively few foreclosures overall, the incidence of foreclosures on non-owner-occupied properties could be low in an absolute sense and yet account for a high share of the state’s few foreclosures. Conversely, in states with many foreclosures, non-owner-occupied properties could account for a 30 Here we use non-occupants’ share, not the change in non-occupants’ share, to focus on the relationship between the level of non-occupant activity and housing appreciation. 31 During the boom years, the highest correlations in Figure 9 result when neither variable is lagged. The lowest correlations result when the variable for housing price appreciation is lagged, but even these are significantly positive in 2005-2006. There is also a slight asymmetry between the two correlations with lagged variables. This asymmetry is somewhat similar to Wheaton and Lee’s findings on lead-lag relationships between sales and prices for total home purchases. See Wheaton and Lee (2008). 32 See Wheaton and Nechayev (2008) for a related finding. They find that the share of non-occupant mortgage borrowing is positively correlated with errors in forecasting housing prices. 33 A key question is whether the LPS data we use accurately capture the regional variation in non-occupant borrowing and foreclosure. Appendix 2 presents evidence that the LPS data are reliable for this purpose. p. 10 of 69 relatively low share of overall foreclosures and yet be much more common and significant than in low-foreclosure states. This is not just a hypothetical issue, in our view, for some of the nonoccupant foreclosure shares in Figure 10 seem to have been materially influenced by the relatively good or bad performance of owner-occupied mortgages in certain states. For example, the map shows that the non-occupant share of foreclosures was 13 percent in New Mexico and 12 percent or less in Arizona. However, foreclosure rates on owner-occupant mortgages originated in 2006 were about three times as high in Arizona as in New Mexico. As a result, using the share of foreclosures to measure the impact of non-occupant borrowing artificially dilutes the scale of the non-occupant foreclosure problem in Arizona and, relatively speaking, exaggerates it in New Mexico. Accordingly, we prefer to use a measure that is not affected by the extent of foreclosures among owner-occupants and reflects only the prevalence and performance of non-occupant mortgages. By prevalence, we mean, for a given year of origination, the number of non-occupant mortgages divided by the total number of housing units.34 By performance, we mean the foreclosure rate (the number of foreclosures on non-occupant mortgages divided by the total number of non-occupant mortgages), again for a given year of origination. The product of these two factors is what we call the impact of non-occupant foreclosures, or the number of foreclosures on non-occupant mortgages divided by the total number of housing units. In other words, Impact = Prevalence x Performance. We make use of this relationship to show that states vary significantly not only in the overall impact of non-occupant foreclosures but also in the two factors, prevalence and performance, that determine the overall impact. Using LPS data on annual cohorts of home-purchase and refinance mortgages, Figure 11a shows how the prevalence and performance of non-occupant mortgages evolved nationally for mortgages originated each year between 2004 and 2007. In 2004, for example, the data include 404 mortgages originated to non-occupant borrowers for every 100,000 housing units in the U.S. Of the mortgages originated to non-occupants in 2004, 4.5 percent had been foreclosed on or were in foreclosure by July 2009, the last month for which we have data. Together, the prevalence and performance of non-occupant mortgages originated in 2004 imply that there were about 18 foreclosures on nonoccupant mortgages originated for every 100,000 housing units. Compared with 2004, the impact of foreclosures on non-occupant mortgages increased for loans originated in 2005 and 2006. This is partly because non-occupant mortgages became more prevalent (533 and 471 per 100,000 housing units, respectively) during this time period and also because their performance deteriorated (to foreclosure rates of 8.5 and 12.7 percent, respectively). As a result, our overall measure of impact (non-occupant foreclosures per 100,000 housing units) rose to 45.1 in 2005 and then to 59.6 in 2006, well above the overall 2004-2007 average of 37.6 foreclosures per 100,000 housing units.35 34 We analyze first-lien, home-purchases plus refinanced mortgages on single-family homes, but we omit homeimprovement loans. The data on state housing units come from the American Community Surveys for the time period 2004–2007. In Appendix 5, we present an alternative measure of mortgage prevalence based on non-occupant mortgage originations as a percentage of total first-lien home-purchases plus refinanced mortgage originations in the same calendar year. Results based on this ―per mortgage‖ measure of the prevalence of non-occupant mortgages are similar to the ―per housing unit‖ results presented in the body of this paper, but the details of state rankings change. We focus on the per housing unit measure because it captures the broadest base for measuring the impact of non-occupant foreclosures and is not affected by state-to-state differences in the intensity of mortgage lending, as noted in a paper written by Mayer and Pence (2008). 35 The full impact of non-owner occupied mortgages is higher than the numbers reported in this study, because the LPS data do not cover the entire mortgage market and may underestimate the share of mortgages to non-occupants. The p. 11 of 69 This can be seen in Figure 11a by noting that 2004 lies below the curved line while 2005 and 2006 lie above, for the line shows all the combinations of prevalence and performance that equate to the average national impact for non-occupant foreclosures during the time period, 2004-2007. The U.S. foreclosure impact for non-occupant mortgages originated in 2007 is 27.4 per 100,000 housing units, which is below the 2004-2007 national average. This was partly due to the housing bust, which reduced the prevalence of non-occupant mortgages sharply to 358 per 100,000 housing units.36 Table 1 summarizes the data in Figure 11a and provides an overview of the impact of nonoccupant foreclosures in 2004-07. Table 1: Prevalence, Performance and Foreclosure Impact of Non-Occupant Mortgages in our LPS Data, 2004-07 Year of Mortgage Origination 2004 2005 2006 Performance (Percent of NonOccupant Mortgages Foreclosed) 4.5 8.5 12.7 Prevalence (Number of NonOccupant Mortgages per 100,000 Housing Units) 404 533 471 Impact (Number of Non-Occupant Mortgage Foreclosures per 100,000 Housing Units) 18.1 45.1 59.6 2007 2004-2007 7.6 8.5 358 441 27.4 37.6 Figures 11b to 11e show how the relative impact of non-occupant foreclosures varied across U.S. states for the years 2004-2007, as well as how the impact outcomes were driven by the underlying prevalence and performance factors. In these figures, prevalence, performance, and impact are measured relative to national norms.37 This means that the point 1.0 on the horizontal axis stands for a level of prevalence equal to the 2004-07 U.S. average for prevalence. Similarly, the point 1.0 on the vertical axis stands for a performance level equal to the 2004-2007 U.S. average for performance. The highest point in Figure 11b is labeled IN for Indiana, and it has an x-axis value of about 0.64. This means that Indiana’s prevalence measure, non-occupant home-purchase and refinance mortgages originated in Indiana in 2004 as a fraction of total housing units in Indiana in 2004, is about 0.64 times the corresponding 2004-07 U.S. average for the prevalence of nonoccupant mortgages. On the y-axis, the value for Indiana is almost 2. This means that Indiana has a foreclosure rate for non-occupant mortgages originated in 2004 that is almost 2 times higher than the corresponding 2004-2007 U.S. average. The product of these two factors for Indiana is about 1.26. This degree of relative impact means that the number of foreclosures on non-occupant degree to which the data underestimate non-occupant foreclosures is discussed in Appendix 2, where we show that the LPS data capture 50 to 60 percent as many mortgage originations as the HMDA data and also imply a share of nonoccupant mortgages in 2006 that is about 27 percent lower than in the HMDA data (10.5 percent versus 14.4 percent). In Appendix 2, we adjust our data to take account of the LPS data’s limited coverage of originations. This raises our impact measures significantly. For example, the peak impact (in 2006) rises to 103.2 non-occupant foreclosures per 100,000 housing units. However, we show in the appendix that the adjusted data and the unadjusted data imply similar relative impacts across the states, and on that basis we use the unadjusted data in the body of this paper. We do not adjust for the difference in non-occupants’ share of mortgage originations between the LPS and HMDA data, because we don’t know which is more accurate. However, if we were to assume that the higher share implied by HMDA is correct, our impact figures would rise to 37 percent in 2006. (This increase is derived by dividing 14.4 by 10.5.) Along with our coverage adjustment, this would imply a non-occupant foreclosure impact of 141.5 per 100.000 housing units for mortgages originated in 2006. 36 Another factor is that loans originated in 2007 have had less time to enter foreclosure than loans in the earlier years. 37 The data for these figures appear in Table A4-1 of Appendix 4. p. 12 of 69 mortgages originated in Indiana in 2004, as a percentage of housing units in Indiana in 2004, was 26 percent higher than the corresponding national average for the period 2004-07. When relative prevalence and relative performance are both 1.0, they multiply to create a relative impact measure of 1.0 as well. A relative impact measure of 1.0 means that non-occupant foreclosures equal the 2004-07 national average of 37.6 per 100,000 housing units. The middle of the three curves sweeping from the upper left toward the lower right in Figures 11b-11e illustrates all points for which the relative impact measure is 1.0. For any state whose dot lies above this curve, the impact of non-occupant foreclosures exceeds the 2004-07 national average impact, and vice versa for points below the curve. The lower-most and upper-most curves in the graph illustrate points where non-occupant foreclosures per housing unit (impact) are, respectively, half of and three times the 2004-07 national average. For mortgages originated in 2004, both Indiana and Nevada have impact measures that clearly exceed the 2004-07 national average. However, in most states, the impact of non-occupant mortgages originated in 2004 was at or below the 2004-07 average. Figure 11b also tells us something about why foreclosures on non-occupant mortgages originated in 2004 are relatively important (or not) in each state. Note that Indiana suffered an above-average impact from non-occupant mortgage foreclosures because of performance problems, not an unusually high prevalence of non-occupant mortgages. On the other extreme, Nevada experienced a relatively low foreclosure rate on mortgages to non-occupant owners but had an above-average impact because non-occupant mortgages were unusually prevalent. Figures 11c, 11d, and 11e present the same analysis but for non-occupant mortgages originated in 2005, 2006, and 2007. From 2004 to 2005, the whole distribution of impact measures shift toward the northeast, as non-occupant mortgages surged to a peak level of prevalence while also performing more poorly. In three states—Nevada, Florida, and Arizona—the impact of nonoccupant foreclosures from 2005 originations reached or exceeded 3 times the 2004-07 national average. This was driven by high prevalence in Arizona and a combination of high prevalence and poor performance in Nevada and Florida. For a cluster of Midwestern states (Indiana, Michigan, Ohio, Minnesota, and Missouri), the foreclosure impact for 2005 non-occupant originations reached or exceeded the 2004-07 national average, even though this type of mortgage was not especially prevalent. However, these states have above-average impact measures due to below-average performance of non-occupant mortgages. Very poor performance caused the foreclosure impact from non-occupant mortgages to peak in 2006 even though prevalence declined that year. Visually this shows up as a shift toward the Northwest in the distribution of impact measures. Several Midwestern states continued to experience very poor performance combined with unexceptional prevalence. Deteriorating performance combined with still high prevalence kept the impact numbers very high in Nevada, Florida, and Arizona, while additional states began to experience impact outcomes that were above the 2004-2007 national average. In the case of Hawaii, Utah, and Idaho, high impact outcomes were driven by high prevalence, while poor performance was the problem in states like Indiana, Michigan, and Ohio. In most states, a combination of lower prevalence and better performance reduced the foreclosure impact of non-occupant mortgages originated in 2007. This is seen in a shift toward the Southwest in the distribution of impacts in Figure 11e, as compared to Figure 11d. However, the level of impact remained quite high in Nevada, Florida, and Arizona. p. 13 of 69 To provide a clearer view of the geographic patterns in non-occupant foreclosure impact and its underlying factors, Figures 12, 13, and 14 use maps to show the prevalence, performance, and impact for 2006 mortgage originations in the 50 states and the District of Columbia. Figure 12 shows that in 2006, non-occupant mortgages were relatively prevalent in the West (including Hawaii); along the mid- to lower-East Coast from Florida to New Jersey; and in Vermont. Since we have already seen that in several of these states the impact of non-occupant mortgages was not especially high, it’s not surprising that Figure 13 shows relatively low non-occupant foreclosure rates in many Western and mid- and lower-East Coast states, the big exceptions being California, Nevada, and Arizona in the West and Florida and Georgia on the lower East Coast. Figure 13 also shows that foreclosure rates for non-occupant mortgages were well above-average in much of the Midwest and parts of the Northeast. Figure 14 presents the effect of prevalence and performance on the relative impact of non-occupant mortgages. States that rank high in both Figures 12 and 13, like Florida and Nevada, are also high (in red) in Figure 14. The figures printed in black indicate that the impact of non-occupant foreclosures was 752 percent of the national average in Nevada and 645 percent of the national average in Florida. The eight states with the highest impacts are all in the West or Southeast. Michigan, Indiana, Rhode Island, Minnesota, Ohio, Illinois, Missouri, and Connecticut are all above average in Figure 14 due to poor performance, or high foreclosure rates for their non-occupant mortgages.38 VI. Conclusion Non-occupant home buyers make up a distinct and significant segment of the U.S. housing and mortgage markets. We show that compared to owner-occupants, non-occupant borrowers tend to have higher incomes, higher credit scores, smaller loans, and generally a lower overall risk profile. Nonetheless, the national rate of foreclosure on mortgages to non-occupants is comparable to the foreclosure rate on mortgages to owner-occupants, presumably due to non-occupants’ greater willingness to use their foreclosure option when the value of housing falls. We also show that, the share of non-occupants borrowing to buy homes was positively correlated with increases in housing prices during the final years of the housing boom. Lastly, we document that the prevalence and performance of non-occupant mortgages varies significantly by state, leading to significant differences in the overall impact of non-occupant foreclosures in local housing markets. In Michigan, Indiana, and some other Midwestern and Northeastern states, the overall incidence of non-occupant foreclosures exceeds the national average mainly due to the poor performance of non-occupant mortgages. By contrast, in Idaho and some other Western states, the relatively high incidence of non-occupant foreclosures is primarily driven by the relatively high prevalence of non-occupant mortgages. The states that experienced the highest impact from foreclosures on properties owned by non-occupants (Florida, Nevada and Arizona) exhibit both relatively poor performance and relatively high prevalence of non-occupant mortgages. 38 Maps covering mortgages originated in 2004, 2005, and 2007 are presented in Appendices Four. p. 14 of 69 Figure 1 Two Facets of the Housing Cycle 18 12 16 10 14 8 12 6 10 4 8 2 6 0 4 -2 2 -4 0 -6 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 Non-Owner-Occupant Mortgage Share (left axis) p. 15 of 69 FHFA Home Price Index (right axis) Percentage Change, April-to-April Percent of First-Lien Home-Purchase Mortgages, by Value (HMDA) (Housing Prices and Mortgages to Non-Occupants) Figure 2: Distributions Across States of the Median Income of Owner-Occupant and Non-OwnerOccupant Mortgage Borrowers, by Loan Purpose and Year of Origination (HMDA) Median Borrower Income by Occupancy (Home Purchase) 150 100 CA CA DC CA HA CA 50 $ Thousands 200 (Box-Whisker Distribution over U.S. States, 2004-07) 2004 2005 owner_occupant 2006 2007 non_occupant_owner (Each box covers the 25th to 75th percentile; the line in the box is the median. ) Median Borrower Income by Occupancy (Refinance) 120 100 CA 80 CA 60 $ Thousands 140 160 (Box-Whisker Distribution over U.S. States, 2004-07) 2004 2005 owner_occupant 2006 2007 non_occupant_owner (From HMDA data; each box covers the 25th to 75th percentile; the line in the box is the median. ) [Note: The data are median applicant income for each state and the District of Columbia, segregated by occupancy. The line in each box is the median of the distribution of median incomes across the 50 states and the District of Columbia. Each box covers the inter-quartile range (25th percentile to the 75th percentile) of the distribution. The ―whiskers‖ extend beyond the box either to the end of the distribution or to a length of 1.5 times the inter-quartile range, whichever comes first. Dots beyond the whiskers identify extreme outlier jurisdictions.] p. 16 of 69 Figure 3a: Distribution Across States of the Mean FICO Score of Owner-Occupant and NonOwner-Occupant Mortgage Borrowers, by Year of Origination (LPS) FICO Score Distributions by Type of Owner 760 (Box-Whisker Distribution over U.S. States, 2004-07) 720 LA AK 680 700 LA MS MS MS MS 660 FICO 740 CO 2004 2005 owner_occupant second_home 2006 2007 investor (From LPS data. Each box covers the 25th to 75th percentile; the line in the box is the median. ) [Note: The data are mean FICO scores for each state and the District of Columbia, segregated by occupancy. Loans whose loan-to-value ratio exceeded 400 were dropped.] p. 17 of 69 Figure 3b: Distribution Across States of Loan-to-Value Ratios for Owner-Occupant and NonOwner-Occupant Mortgage Borrowers, by Year of Origination (LPS) Loan-to-Value Distributions by Type of Owner 80 (Box-Whisker Distribution over U.S. States, 2004-07) 65 70 75 AK CA HA 60 CA RI MA 2004 2005 owner_occupant second_home 2006 2007 investor (From LPS data. Each box covers the 25th to 75th percentile; the line in the box is the median. ) Note: Loan-to-value ratios here are computed as the original principal of the mortgage divided by the most recently available appraised value of the mortgaged property. Loans whose loan-to-value ratio exceeded 400 were dropped. p. 18 of 69 Figure 3c: Distribution Across States of Debt-to-Income Ratios of Owner-Occupant and NonOwner-Occupant Mortgage Borrowers, by Year of Origination (LPS) Debt to Income Distributions by Type of Owner SD SD ND NE AR ND NE SD VT 40 ND OH 30 DTI 50 60 (Box-Whisker Distribution over U.S. States, 2004-07) AR NY NY NY 20 NY DC 2004 2005 owner_occupant second_home 2006 2007 investor (From LPS data. Each box covers the 25th to 75th percentile; the line in the box is the median. ) Note: Loans whose loan-to-value ratio exceeded 400 were dropped. p. 19 of 69 Figure 4: Distribution Across States of the Median Amount of Owner-Occupant and Non-OwnerOccupant Mortgages, by Loan Purpose and Year of Origination (HMDA) Median Mortgage Amount by Occupancy (Home Purchase) 400 (Box-Whisker Distribution over U.S. States, 2004-07) 300 HA 200 CA HA 0 100 $ Thousands CA HA CA HA CA HA DC 2004 2005 2006 owner_occupant 2007 non_occupant_owner (From HMDA data; each box covers the 25th to 75th percentile; the line in the box is the median. ) Median Mortgage Amount by Occupancy (Refinance) 400 (Box-Whisker Distribution over U.S. States, 2004-07) CA HA 200 0 100 $ Thousands 300 CA HA 2004 2005 owner_occupant 2006 2007 non_occupant_owner (From HMDA data; each box covers the 25th to 75th percentile; the line in the box is the median. ) [Note: The data are median mortgage amount for each state and the District of Columbia, segregated by occupancy.] p. 20 of 69 Figure 5: A Cross-State Comparison of Foreclosure Rates on Non-Owner-Occupied and OwnerOccupied Mortgages, by Year of Origination (2004-2007) Home Purchase and Refinance Originations, LPS FL MI MN 10 MS GA MOKY IL WV AZ RI AL TN NY CT OK NJ WI LAMATX IA NHSC CO PA DCHA MD KSNEME ND AKSD NCAR VA ID DE UTNM OR WA VT MT WY MO MN TN KY CO 5 Owner-Occupied Foreclosure Rate (%) 10 0 2006 Foreclosure Rates by Occupancy 20 15 RI 15 FL IN OH MN GA 5 10 Owner-Occupied Foreclosure Rate (%) NV MI MI NV AZ CA AZ 10 Non-Owner-Occupied Foreclosure Rate (%) 25 20 15 5 CT NJ MS MO NY MD TN AL KY DC WI OK WVVA MA PA TXHA ID SC NHLA AK KSAR DE IACO NE UT OR NCWA ME SDNM VT MT MN OHIN GA MS IL MD NJ UT MO ID WI CT TN ALSC NHNY DCLAWV OR VA KY NMHA WA PAMA AR CO ME OK TX KS DE IA AK NC NEVT MT SD WY ND 0 10 5 0 5 CA Home Purchase and Refinance Originations, LPS FL IL 0 GA MS 2007 Foreclosure Rates by Occupancy Home Purchase and Refinance Originations, LPS WY ND NV 0 5 0 OH 15 OH IL IAAL NE OK NY NV LA TX RI FLKS WI WV SDPANH NJCT NC MA SC NDCA AKUT ME AR DCMD AZ WY ID NM VT DE WAMT HA VA OR 0 MI IN IN 5 10 15 Non-Owner-Occupied Foreclosure Rate (%) 20 2005 Foreclosure Rates by Occupancy Home Purchase and Refinance Originations, LPS 20 2004 Foreclosure Rates by Occupancy 10 15 20 Owner-Occupied Foreclosure Rate (%) 25 p. 21 of 69 0 CA RI 5 10 Owner-Occupied Foreclosure Rate (%) 15 Figure 6 Home Sales by Use, 2003-07 (from the National Assn. of Realtors) 9 8 7 Million Units 6 5 4 3 2 1 0 2003 2004 2005 Primary Residence p. 22 of 69 Vacation 2006 Investment 2007 Figure 7 Home-Purchase Mortgage Originations by Occupancy (HMDA data on number of originations) 8 7 6 Millions 5 4 3 2 1 0 1996 1997 1998 1999 2000 Owner-Occupied p. 23 of 69 2001 2002 2003 Non-Owner-Occupied 2004 2005 2006 Figure 8 Relative Change in House Prices and Non-Occupant Share of Mortgages 350 HMDA Purchase Mortgages and FHFA Housing Price Index, 1996-2006 GA MN ID 300 NM DE 250 PAMT 200 TX MI 150 OH WY TN UT AL SCIL NC WVMOAK CO KSND LA IN IAAR KY OK MS WI SD VT DC NY MD CT HAVA NJ WA NH ME AZ NV OR MA FL RI CA 100 NE 100 150 200 250 300 % Change in Housing Prices 1996-2006 p. 24 of 69 350 Figure 9 0.60 Cross-State Correlations Between Non-Occupants' Share of HomePurchase Mortgage Originations and Annual Percentage Changes in Housing Prices, 1997-2006 0.50 0.40 0.30 0.20 0.10 0.00 -0.10 -0.20 -0.30 -0.40 1997 1998 No Lags 1999 2000 2001 Mortgage Share Lagged 1 Year p. 25 of 69 2002 2003 2004 2005 Price Change Lagged 1 Year 2006 Figure 10: Share of Foreclosure That Involve Non-Owner-Occupied Properties (LPS data for mortgages originated in 2006) p. 26 of 69 Figure 11a: National Prevalence and Performance of Non-Occupant Mortgages, by Year of Origination, 2004-2007 U.S. Non-Occupant Mortgage and Foreclosure Impact 2004-07 Home Purchase and Refi Originations (LPS) 10 12 2006 8 2005 6 2007 4 2004 300 400 500 Prevalence (Non-Occupant Mortgages per 100,000 Housing Units) 600 2004-07 Natl. Avg.= 37.6 foreclosures per 100,000 housing units p. 27 of 69 Figures 11b, c, d, and e: The Relative Impact (per Housing Unit) of Non-Occupant Foreclosures, by States and Underlying Factors, for Mortgages Originated in 2004, 2005, 2006, and 2007 Relative Non-Occupant Foreclosure Impact per Housing Unit 2 3 2004 Home Purchase and Refinance Originations (LPS) IN OH MI MS NV HA 0 1 GA MO KY TN MN IL IA AL NE NY OK KS TX RI FL WVLA PA SDWI CO CT MA NJ NCNH SC AR UT ND AK ME MD CA DC ID WY NM VT VA MT WA DE AZ OR 0 1 2 3 Relative Non-Occupant Mortgage Prevalence (U.S.=1) 4 Relative Non-Occupant Foreclosure Impact per Housing Unit 3 2005 Home Purchase and Refinance Originations (LPS) 2 MI IN OH NV FL AZ DE UT ID HA 0 1 GA MN MS MO KY IL WV CA RI AL TN NY CT OKWI NJ IA LA TX MANH CO SC NE KS PA ME VA MD DC NDSD AK NC AR NM OR VT WA MT WY 0 1 2 3 Relative Non-Occupant Mortgage Prevalence (U.S.=1) p. 28 of 69 4 Relative Non-Occupant Foreclosure Impact per Housing Unit 3 2006 Home Purchase and Refinance Originations (LPS) FL NV MI 2 IN OH MN RI GA AZ CA ID HA UT 0 1 IL CT MS NJ MO NY MD TN AL KY WI DC VA WVLA OKPA MA SC TX NH CO AK DE AR IA NE KS OR ME NC WA NM SD VT MT WY ND 0 1 2 3 Relative Non-Occupant Mortgage Prevalence (U.S.=1) 4 Relative Non-Occupant Foreclosure Impact per Housing Unit 3 2007 Home Purchase and Refinance Originations (LPS) 2 FL MI NV AZ 0 1 MN OH GA IN CA MS IL NJ MD WI RIMO TN NH WVLA AL NY CT SC DC KY OR VA NM WA MA PA AR CO OK TX MENC KS DE IA AK VT NE MT SD WY ND 0 ID UT HA 1 2 3 Relative Non-Occupant Mortgage Prevalence (U.S.=1) 4 Note: Middle line represents an impact equal to the LPS data 2004-07 national average of 37.6 non-occupant foreclosures per 100,000 housing units. The lower and upper lines represent impacts of, respectively, half and three times that national average impact. p. 29 of 69 Figure 12: Non-Owner-Occupied Mortgage Prevalence per Housing Unit, Relative to 2004-2007 U.S. Average (LPS data for mortgages originated in 2006) p. 30 of 69 Figure 13: Non-Owner-Occupied Mortgage Foreclosure Rates, Relative to 2004-2007 U.S. Average (LPS data for mortgages originated in 2006) p. 31 of 69 Figure 14: Non-Owner-Occupied Mortgage Foreclosure Impact per Housing Unit, Relative to U.S. 2004-2007 Average (LPS data for mortgages originated in 2006) p. 32 of 69 References Apgar, William C., and Shekar Narasimhan (2008). ―Capital for Small Rental Properties: Preserving a Vital Housing Resource.‖ Chapter 8 in: Revisiting Rental Housing: Policies, Programs, and Priorities, ed. Nicolas P. Retsinas and Eric S. Belsky. Brookings Institution Press, Washington, DC. Avery, Robert B., Kenneth P. Brevoort, and Glenn B. Canner (2007). ―The 2006 HMDA Data.‖ The Federal Reserve Bulletin 93(Dec.): A73–A109. —— (2009). ―Geographic Patterns in Mortgage Loan Performance.‖ Unpublished working paper, Board of Governors of the Federal Reserve System. Bajari, Patrick, Chenghuan Sean Chu, and Minjung Park (2008). ―An Empirical Model of Subprime Mortgage Default from 2000 to 2007.‖ NBER Working Paper 14625. National Bureau of Economic Research. Benjamin, John D., Peter Chinloy, William G. Hardin III, and Zhonghua Wu (2008). ―Clientele Effects and Condo Conversions.‖ Real Estate Economics 36(3): 611–634. Coulton, Claudia, Tsui Chan, Michael Schramm, and Kristen Mikelbank (2008). Pathways to Foreclosure: A Longitudinal Study of Mortgage Loans, Cleveland and Cuyahoga County, 2005– 2008. Report of the Center on Urban Poverty and Community Development, Mandel School of Applied Social Sciences (June). Case Western Reserve University, Cleveland, OH. Cowan, Adrian M., and Charles D. Cowan (2004). ―Default Correlation: An Empirical Investigation of a Subprime Lender.‖ Journal of Banking and Finance 28: 753–771. Gerardi, Kristopher, Adam Shapiro, and Paul Willen (2008). ―Subprime Outcomes: Risky Mortgages, Homeownership Experiences, and Foreclosures.‖ Working Paper 07-15 (May 4, 2008, version). Federal Reserve Bank of Boston. Haughwout, Andrew, Richard Peach, and Joseph Tracy (2008). ―Juvenile Delinquent Mortgages: Bad Credit or Bad Economy?‖ Staff Report 341 (August). Federal Reserve Bank of New York. Immergluck, Dan, and Geoff Smith (2004). Risky Business—An Econometric Analysis of the Relationship Between Subprime Lending and Neighborhood Foreclosures. Woodstock Institute, Chicago. Mayer, Chris, and Karen Pence (2008). ―Subprime Mortgages: What, Where, and to Whom?‖ Finance and Economics Discussion Series paper 2008-29. Board of Governors of the Federal Reserve System. McCormick, Erin, and Carolyn Said (2007). ―Investors Own About One-Fifth of Bay Area Homes in Foreclosure.‖ San Francisco Chronicle (Dec. 16). p. 33 of 69 Qi, Min, and Xialong Yang (2007). ―Loss Given Default of High Loan-to-Value Residential Mortgages.‖ Economics and Policy Analysis Working Paper 2007-4 (August). Office of the Comptroller of the Currency, U.S. Treasury. Todd, R.M. (2010). ―Foreclosures on Non-Owner-Occupied Properties in Ohio’s Cuyahoga County: Evidence from Mortgages Originated in 2005-2006.‖ Community Affairs Research Report 2010-2. Federal Reserve Bank of Minneapolis. Wheaton, William C., and Nai Jia Lee (2008). ―Do Housing Sales Drive Prices or the Converse?‖ Working Paper 08-01 (Jan. 28). Department of Economics, Massachusetts Institute of Technology, Cambridge. p. 34 of 69 Appendix 1: A simple theory of investor mortgage default The mortgage payment decision for both non-occupants and owner-occupants is affected by the option to default on the mortgage. We modify the model presented in Gerardi, Shapiro, and Willen (2008) by assuming that non-occupants (―investors‖) are purely profit-maximizers. Our model is a simple, two-period analysis. In period zero, the non-occupant owner owns the property and has financed it with an interest-only mortgage.39 In period one, the mortgage is due. Refinancing the mortgage is not an available option in the model. As a result, the owner in period zero has the option of making the mortgage payment and keeping the property, selling the property to pay off the mortgage, or defaulting on the mortgage. The payoff structure under each scenario can be written as follows: Option 1: Pay the mortgage in period 0; sell the property in period 1. π0 = ρ - rMU π1 = (H1 – U)+, where π0 is profit in period zero, π1 is profit in period one, ρ is net rental income, rM is the mortgage interest rate, U is the unpaid balance on the investor’s interest-only mortgage, H1 is the price of the house in period one, and (H1 – U)+ means either (H1 – U) or zero, whichever is greater.40 This last term reflects the fact that, in period one, the investor will sell the house and pay off the mortgage if H1 ≥ U but will simply default if not. Option 2: Sell the property in period 0. π0 = H0 – U π1 = 0, where H0 is the price of the house in period zero. Option 3: Default on the mortgage in period 0. π0 = π1 = 0. Options 2 and 3 can be combined into a single ―Don’t keep the property‖ option in which π0 = (H0 – U)+ and π1 = 0. Then the entire investor decision in this simple framework boils down to keeping the property (making the period one mortgage payment) if ρ + {[1/(1+r)](H1 – U)+ - (H0 – U)+} ≥ rMU, 39 Before purchasing the property, and in particular before making the down payment, the investor would analyze capital gains by comparing future property values to the purchase price (albeit while also recognizing that potential losses are limited to the down payment). Thereafter, as in the analysis presented here, the mortgage default option leads the investor to analyze capital gains by comparing the future value of the property to the unpaid balance on the mortgage instead of the current value of the property. 40 We parallel Gerardi, Shapiro, and Willen (2008) also in assuming that housing prices in both periods are known in period zero. As they note, allowing the period one housing price to be unknown as of period zero would mean that both the expected value and the variance of the period one housing price would affect the investor’s decision. p. 35 of 69 where r is the interest rate at which the investor discounts period one profits. The first term on the left is period-zero net rental income. The bracketed term on the left is the present value of expected capital gains from future sale, taking into account the value of the option to default today or tomorrow. The term on the right is the period-zero interest payment on the mortgage. p. 36 of 69 Appendix 2: Comparing the LPS Mortgage Data to Some Alternatives We rely heavily on LPS’s data that are reported by a group of large mortgages servicers. Here we assess and adjust for how the LPS data differ from other national and local data on mortgages, focusing on first-lien mortgages originated in 2004–2007 for the purposes of either home purchase or mortgage refinance.41 Compared to the HMDA data, the LPS data cover about 50 percent of the mortgages originated in 2004 and about 60 percent of those originated in 2005-07.42 This coverage difference implies that our prevalence and foreclosure impact measures for non-occupant mortgages underestimate the actual levels of prevalence and impact. We provide adjustments that scale up our LPS measures to more realistic levels, and we show that these adjustments do not greatly alter the relative regional distributions of prevalence, performance, and impact discussed in the body of the paper. Similarly, we find that the share of mortgages made to non-occupant owners is lower in the LPS data than in the HMDA data. We show that this difference is fairly uniform across states and should have a minor effect on the relative regional patterns discussed in the body of the paper. Finally, we benchmark the LPS data for selected ZIP-code areas against local records on non-owner-occupied mortgages and foreclosures and again find a high spatial correlation. We conclude that the LPS data are likely to undercount the extent of non-owner-occupant home buying, mortgage borrowing, and mortgage foreclosures, but this undercounting appears to be relatively uniform across the country. As a result, we assert that the LPS data correctly identifies significant regional patterns in non-occupant mortgage prevalence, performance, and impact. Figure A2-1 documents that the LPS data cover fewer mortgages than the HMDA data. The figure compares the total number of mortgage originations43 in the LPS data with the total number of mortgage originations reported under HMDA for each state by year of origination for the period 2004-2007. LPS’s coverage is highest for the most recent year, probably in part because the market share of the servicers reporting to LPS has increased. For 2004, Figure A2-1 shows that mortgage originations in the LPS database are nearly half of the total number of HMDA originations in a typical state, where the fraction of mortgage originations falls within the range of 0.30 to 0.62.44 For home-purchase mortgages originated in 2005-07, LPS’s coverage rises to about 60 percent of the HMDA originations in a typical state, but the fraction falls into the high 30s in some states, while rising to about 80 percent in others. LPS’s incomplete coverage of the mortgage market depresses our estimates of the prevalence and foreclosure impact of non-occupant mortgages. In this appendix, we created adjusted estimates by inflating each state’s annual LPS data by the state’s ratio of HMDA to LPS originations for that year. For example, in Alabama in 2006, the number of mortgage originations reported by LPS was 41 We also limit our comparisons to mortgages for which LPS reports that the borrower is an owner-occupant, a secondhome owner, or an investor. We omit LPS mortgages for which the occupancy status of the borrower is not known to be one of these three types. 42 Some of the LPS-HMDA gap arises because LPS obtains its data from mortgage servicers whose collective market share is less than 100 percent of the total market for mortgage originations. It also reflects our omission of LPS mortgages with unknown owner occupancy type. 43 In the comparison of the LPS and HMDA data in this appendix, we use ―mortgage‖ to refer to loans made to purchase a house or refinance an existing mortgage but exclude home improvement loans. 44 Differences among the states can arise because of regional variations in the speed with which mortgages are paid off or are foreclosed as well as regional variations in the coverage of both the LPS and HMDA data. The collective market share of the mortgage servicers that provide data to LPS may vary regionally, and HMDA’s coverage can be spotty to varying degrees in rural areas, for example. p. 37 of 69 about 56.1 percent of the corresponding figure from HMDA. Thus, for our adjusted estimates, we scale the LPS mortgage data for Alabama for 2006 by 1.78 (the inverse of 0.561) and make corresponding adjustments for the other states and years. These adjustments significantly raise the level of all of our prevalence-per-housing-unit and impact-per-housing-unit measures.45 For example, the peak national non-occupant foreclosure impact based on just the LPS data was 59.6 non-occupant foreclosures per 100,000 housing units, for mortgages originated in 2006 (as discussed in conjunction with Figure 11 above.). After scaling up the LPS data, we compute that the impact measure is 73 percent higher, or 103.2 non-occupant foreclosures per 100,000 housing units for mortgages originated in 2006. It is obvious that scaling up the LPS mortgage data to match HMDA’s state-by-state total of mortgage originations significantly raises the level of our per-housing-unit prevalence and foreclosure impact measures. However, we find that it does not greatly change the cross-state spatial distribution of prevalence and impact that was discussed above in conjunction with Figures 11, 12, and 14. For example, across the 50 states and the District of Columbia in the year of peak impact (2006), the correlation coefficient for the unadjusted and adjusted impact measures is 0.99. The general similarity in regional patterns can be seen by comparing the adjusted data shown in Figures A2-2a through A2-2d with the unadjusted data shown in Figures 11b through 11e. Finally, Table A2-1 provides a complete listing of our adjusted measures for 2004-2007 originations, for comparison with the unadjusted data presented in Table A4-1 below. Although the two tables differ in detail, states that rank high (low) in the unadjusted data also rank high (low) in the adjusted data. The LPS data also show a lower share of mortgages to non-occupants, compared to the HMDA data. In the LPS data, the percentage of mortgages to non-occupants was about 10 percent in 20042006, or about three to four percentage points less than the corresponding shares computed from HMDA originations.46 It is not clear whether the LPS or HMDA fraction is more accurate, as both are subject to misreporting and less than complete coverage. For that reason, we make no detailed adjustments for this difference. However, if we were to assume that the HMDA figures are correct, then the low share of non-occupants in the LPS data would imply about a 25 percent underestimate of the prevalence and foreclosure impact of mortgages to non-occupant owners. Although the share of mortgages to non-occupants could be biased downward in the LPS data, it seems to vary among the states much like the corresponding HMDA shares. Figures A2-3a to A23d compare LPS and HMDA data for 2004-07 across states with regard to the share of mortgages (home-purchase and refinance) to non-occupant owners. The percentages are generally lower in the LPS data, as discussed above and as indicated by the fact that most of the points in these figures are on or near the 45-degree line, which depicts equal shares between the two data sets. Nonetheless, the LPS and HMDA shares generally move together, so that states that are relatively high by one measure are also relatively high according to the other. This is confirmed by high statistical correlations—0.87 for 2004, 0.91 for 2005, 0.88 for 2006, and 0.83 for 2007—between the two measures of the share of mortgages to non-owner occupants. Overall, we find that the relative regional patterns for non-occupant mortgages and foreclosures are similar in the LPS and HMDA data. 45 All of our performance measures and the per-mortgage measures of prevalence and impact that are found in Appendix 5 are only slightly affected, because the adjustment affects the numerators and denominators by similar amounts. 46 More precisely, the LPS data put the non-occupant share of home-purchase plus refinance originations at 8.7 percent in 2004, 10.2 percent in 2005, 10.5 percent in 2006, and 10.3 percent in 2007. The corresponding shares based on HMDA originations are 11.9, 13.5, 14.4, and 14.1 percent, respectively. p. 38 of 69 We have also compared the LPS data at the ZIP-code level with data from local property records in Ohio and Minnesota. Our most extensive ZIP-level comparison involves 49 ZIP Code areas in Cuyahoga County, Ohio, where we obtained local data on mortgages originated in 2005 and 2006 from Michael Schramm of Case Western Reserve University.47 The Case Western data are estimated to cover about 68 percent of all local originations. LPS’s coverage is again lower; the number of mortgages in the LPS data equals about 55 percent of the number of mortgages in the Case Western data for 2005 and about 61 percent for 2006. Despite this difference in coverage, the share of mortgages to non-occupants in the LPS data for these ZIP Codes matches fairly well with the share computed from the Case Western data. Among mortgages originated in 2005 in the 49 ZIP Code areas, the LPS data show an 11.5 percent non-occupant share, compared to a 13.2 percent share in the Case Western data. For 2006, the share computed from the LPS data is 14.5 percent, slightly lower than the 14.8 percent share computed from the Case Western data. As shown in Table A2-2, the Case Western data imply somewhat lower foreclosure rates than what we find in the LPS data. These differences partly reflect the time periods involved. The Case Western data tabulate whether a mortgage has had any foreclosure notice between its origination date and April 2008. The LPS data tabulate whether a mortgage has been in the foreclosure process (―pre-sale‖ status) between origination and July 2009, or 15 months longer than the period tracked by Case Western. Note that both data sources imply that the rate of foreclosure on non-owneroccupied mortgages was about 1.7 to over 3 times higher than on owner-occupied mortgages. Despite their different levels of foreclosure, the two sources agree on how lending and foreclosure rates varied among the 49 ZIP Codes, as shown in Table A2-3.48 This reinforces the conclusion that the LPS data provides reliable information about the spatial pattern of non-occupant borrowing and foreclosure. We also compared the LPS data to local records for two ZIP Codes in Ramsey County, Minnesota, which includes the city of St. Paul. Minnesota provides special property tax relief to owneroccupants. To receive the credit, owner-occupants must file for the state’s homestead credit. County records show which properties are on file as owner-occupied homesteads. Through a laborintensive process, staff at the Federal Reserve Bank of Minneapolis pulled mortgage records for samples of both homesteaded and non-homesteaded single-family residences in ZIP Code areas 55106 and 55109.49 As a result, we can estimate the number of single-family homes in these ZIP Codes that had active mortgages in early 2008. Furthermore, we can estimate the percentage of these mortgaged single-family homes that were non-homesteaded, which serves as an estimate of the percentage of mortgaged single-family homes owned by non-occupants. 47 As before, these mortgage data are for loans made to purchase a house or refinance an existing mortgage and exclude home improvement loans. Collection of the Cuyahoga County data is described in Coulton, Chan, Schramm, and Mikelbank (2008). Using the same data for the years 2005-2006, Todd (2010) shows a pattern of relatively high prevalence, poor performance, and high foreclosure impact associated with mortgages by non-local banks to non-owner occupants in low-income, minority neighborhoods with low housing values. 48 In both 2005 and 2006, one ZIP Code had no non-owner-occupied mortgages and was thus omitted when the correlation for foreclosure rates was computed. For 2005, the LPS data had no non-owner-occupied mortgages in ZIP Code area 44114, and in 2006 the Case Western data had none in ZIP Code area 44115. 49 ZIP Code area 55106 covers much of St. Paul’s ―East Side‖, an inner-city low- to moderate-income area that has recently experienced a high level of foreclosure. ZIP Code area 55109 covers some moderate- to medium-income inner-ring suburbs on the northern and eastern edges of St. Paul. p. 39 of 69 Using these estimates, we find that the share of mortgages to non-occupant owners computed from Ramsey county property records is close to the share computed from the LPS data for our two ZIP Code areas. We estimate that 80 percent of the single-family homes in the ZIP Code area 55106 had active mortgages in early 2008 and that 10.2 percent of them were not homesteaded. Given that we sampled 10 percent of the single-family property records for this ZIP Code, a 95 percent confidence interval would state that the percentage of active mortgages to non-occupants falls in the range of 6.2 to 14.2 percent. The LPS data on 2004-2006 originations imply that 13.1 percent of active mortgages in ZIP Code area 55106 were to non-occupant owners as of March 2008. This percentage is higher than what the property records imply but within the 95 percent confidence interval. In ZIP Code area 55109, about 76 percent of the single-family properties we sampled had an active mortgage in early 2008, and just 3.6 percent of these mortgages were on non-homesteaded properties. In the LPS data, 4.2 percent of the active mortgages in ZIP Code area 55109 are to nonoccupants, a figure which is again higher than what we estimate from property records but within the 95 percent confidence interval. These two Minnesota ZIP Code areas provide a bit more evidence that our LPS data reflect actual regional differences in lending to non-owner-occupants as well as, perhaps, the actual share of this form of lending. p. 40 of 69 Table A2-1: Adjusted Indices of Non-Occupant Mortgage Performance, Prevalence, and Impact Adjusted Indices of the Relative Performance, Prevalence, and Impact of Non-Owner-Occupied (NOO) Mortgages and Foreclosures, by State and Year Mortgage Was Originated (based on LPS data scaled up to each state's number of HMDA originations) Performance Index* Prevalence Index Impact Index (NOO (Foreclosures per mortgage, for NOO (Mortgages per housing unit, for NOO foreclosures per housing unit, by year of mortgages, by year of origination, as a mortgages, by year of origination, as a origination, as a percentage of the 2004-07 percentage of the 2004-07 national average) percentage of the 2004-07 national average) national average) State 2004 2005 2006 2007 2004 2005 2006 2007 2004 2005 2006 2007 88 103 61 68 74 63 60 77 39 AL 68 61 42 AK 30 48 74 32 66 67 56 48 20 32 41 15 AZ 24 99 170 136 243 334 221 149 57 331 376 203 AR 33 44 70 44 48 55 54 50 16 24 38 22 CA 28 100 166 94 153 155 108 83 43 156 179 77 CO 42 62 74 42 148 147 132 124 63 91 98 52 CT 42 81 136 65 73 77 61 46 30 62 83 30 DE 18 40 74 34 192 209 155 123 34 83 114 42 DC 27 54 98 59 140 156 134 92 37 84 131 54 FL 52 160 289 210 230 309 234 136 120 494 677 285 GA 108 136 166 103 95 119 118 86 102 161 195 88 HA 15 54 94 52 298 308 210 143 46 165 197 73 ID 22 41 88 73 182 271 209 146 41 112 184 107 IL 75 107 145 88 62 75 73 59 46 80 106 51 IN 197 207 226 100 74 76 74 54 145 157 168 54 IA 69 68 70 34 37 37 34 31 25 25 23 10 KS 56 57 68 39 58 60 53 46 32 35 36 18 KY 88 111 103 54 50 53 50 42 44 59 52 22 LA 55 69 87 57 54 56 51 48 29 38 44 27 ME 30 55 60 42 115 112 92 72 35 61 55 30 MD 27 56 109 83 129 148 120 83 34 83 131 69 MA 38 65 89 47 91 93 69 57 35 61 62 27 MI 153 210 250 168 72 74 68 54 110 154 169 91 MN 86 128 173 116 73 76 63 48 63 98 110 55 MS 118 120 129 89 47 50 53 55 55 60 68 49 MO 96 115 118 78 80 87 85 72 76 100 101 56 MT 19 13 38 27 89 96 89 82 17 13 33 22 NE 62 56 69 27 36 35 41 37 22 20 28 10 NV 55 170 251 164 379 417 281 162 210 708 703 266 NH 45 60 82 59 105 96 79 62 47 58 65 37 NJ 42 76 125 86 99 114 88 63 42 86 110 55 NM 21 33 57 51 103 138 129 91 22 46 73 46 NY 58 84 115 61 45 50 44 35 26 42 50 21 NC 41 46 61 37 113 136 133 107 46 63 81 39 ND 31 49 27 6 34 34 30 27 11 16 8 2 OH 170 188 216 105 68 76 70 53 117 143 151 55 OK 58 78 96 40 50 55 54 45 29 43 52 18 OR 14 30 66 54 131 161 127 102 19 49 84 55 PA 49 60 89 45 70 80 71 53 35 48 63 24 RI 53 89 159 74 141 128 86 62 74 114 137 46 SC 33 62 85 60 160 209 176 130 53 130 149 78 SD 47 46 55 20 44 47 47 50 21 21 26 10 TN 83 88 108 64 63 78 84 74 53 69 91 47 TX 55 66 84 39 62 70 78 63 34 46 66 24 UT 32 35 67 78 139 187 193 152 44 66 128 119 VT 20 23 44 32 161 146 115 95 32 33 51 30 VA 20 52 91 53 115 131 101 75 23 68 92 40 WA 18 28 61 48 102 120 111 98 18 33 68 48 WV 51 100 91 61 50 54 45 38 26 54 41 24 WI 51 73 100 73 88 89 78 65 45 66 78 48 WY 22 12 29 11 84 78 75 65 19 9 22 7 National 825 per 955 per 799 per 598 per 37.7 per 82.9 per 103.2 per 46.9 per Average 4.57% 8.68% 12.91% 7.84% 100,000 100,000 100,000 100,000 100,000 100,000 100,000 100,000 * Not affected by the adjustment p. 41 of 69 Table A2-2: Case Western and LPS Foreclosure Rates in Cuyahoga County (percent) Owner-Occupied Non-Owner-Occupied Combined 2005 Case Western 11 31 13 2005 LPS 15 25 16 2006 Case Western 8 25 10 2006 LPS 17 33 19 Table A2-3: Correlations between Selected Case Western and LPS Variables 2005 Number of Owner-Occupied Mortgages 0.93 Number of Non-Owner-Occupied Mortgages 0.93 Number of Foreclosures on Owner-Occupied Mortgages 0.96 Number of Foreclosures on Non-Owner-Occupied Mortgages 0.95 Percent of Mortgages Not Owner-Occupied 0.82 Foreclosure Rate on Non-Owner-Occupied Mortgages 0.89 2006 0.91 0.98 0.96 0.97 0.92 0.94 Figure A2-1 LPS vs. HMDA Originations, Home Purchase plus Refi .3 .4 .5 .6 .7 .8 (Box-Whisker Distribution over U.S. States, 2004-07) _2004 _2006 _2005 _2007 LPS data as of Nov. 2009. Boxes cover 25th to 75th percentile; the line in the box is the median. p. 42 of 69 Figure A2-2a (2004, adjusted LPS data) Figure A2-2b (2005, adjusted LPS data) MI IN 2 2 3 2005 Home Purchase and Refinance Originations (Adjusted LPS) Relative Non-Occupant Foreclosure Rate (U.S.=1) Relative Non-Occupant Foreclosure Impact per Housing Unit 2004 Home Purchase and Refinance Originations (Adjusted LPS) 3 Relative Non-Occupant Foreclosure Impact per Housing Unit IN OH OH MI 0 0 GA MN MS KY IL MO WV CA RI ALTN NY OK CTWI NJ IA LA TX MA SC MDDC NE KS PA NH ME VA CO NDSDARAK NC DE OR UT WA NM VT WY MT 1 1 GA MO KYTNMN IL IA AL NENY OK KS LA TXPA WI RI WV SD CO CT MANJNHNC NDAR AK MEMDUT CASC DC WY ID NMVA MTWA OR VT DE NV FL AZ HA AZ ID HA 0 MS NV FL 1 2 3 Relative Non-Occupant Mortgage Prevalence (U.S.=1) 4 Figure A2-2c (2006, adjusted LPS data) 0 1 2 3 Relative Non-Occupant Mortgage Prevalence (U.S.=1) 4 Figure A2-2d (2007, adjusted LPS data) FL MI NV IN OH MN 1 2 3 Relative Non-Occupant Mortgage Prevalence (U.S.=1) MI NV AZ 1 MN OH GA IN MS ILNJ CA MD RI IDUT WIMO CT NH TN WV NY LA AL VA DCOR SCHA KY NM MA ME WA CO PA AR OK TX NC DE IA KS NEAK MTVT SD WY ND 0 0 1 AZ CAGA RI IL CT MS NJ MO NY MD TN KY AL WI DC OK HA WV PATX VA LA MA SC ID NH CO AK DE AR IANEKS UT OR NC ME WA NM SD VT MT WY ND 0 FL 2 2 3 2007 Home Purchase and Refinance Originations (Adjusted LPS) Relative Non-Occupant Foreclosure Rate (U.S.=1) Relative Non-Occupant Foreclosure Impact per Housing Unit 2006 Home Purchase and Refinance Originations (Adjusted LPS) 3 Relative Non-Occupant Foreclosure Impact per Housing Unit 4 p. 43 of 69 0 1 2 3 Relative Non-Occupant Mortgage Prevalence (U.S.=1) 4 Figure A2-3a Figure A2-3b Non-Occupants' Share of Mortgage Originations Non-Occupants' Share of Mortgage Originations HMDA versus LPS, Home Purchase plus Refinance, 2005 IDAZ DE SC .2 VT SC NV FL ME ORMT UTDC CO GA WYWV AL RI TX PA NJVA WA LATNMOOK MS AR IN OH WI AK MD NHCA MA NY KS KY MN MI SD CTND IL IA NE DENM NCAZ ID FL NV 0 0 .1 MENM NC DC ORMT WY UT CO RI WV AL GA AK NHVA NJ TX MD PA CA WA WI IN LA AR MS MO TN OK MA KS NDMI OH SD KY MNNY CT IL NE IA HA HA .1 VT Non-Occupants' Share, LPS .2 .3 .3 HMDA versus LPS, Home Purchase plus Refinance, 2004 0 .1 .2 Non-Occupants' Share, HMDA 0 .3 .1 .2 Non-Occupants' Share, HMDA .3 HMDA=LPS HMDA=LPS Figure A2-3c Figure A2-3d Non-Occupants' Share of Mortgage Originations Non-Occupants' Share of Mortgage Originations HMDA versus LPS, Home Purchase plus Refinance, 2006 VT HA SC .2 Non-Occupants' Share, LPS .2 .3 .3 HMDA versus LPS, Home Purchase plus Refinance, 2007 .1 VT CT HA NV FL SC ID AZ DE CO NC MEMT NM OR UT DC GA TXALTN MS OH MI MO WA WY CA SD NH AR RI VA PA OK INWVWI LA MD MA NY NJ KY AK MN KS ND ILNE IA 0 0 .1 ID FL NV NC DENM ME UT AZMT DC OR CO GA TX AL TN WY IN MO MSAR PA OH WA WI RINH VA MIWV KY OK NJ MD CA NY AK SD LA MA KS NDMN CT IL NE IA 0 .1 .2 Non-Occupants' Share, HMDA .3 HMDA=LPS 0 .1 .2 Non-Occupants' Share, HMDA HMDA=LPS p. 44 of 69 .3 Appendix 3: Investor and Second-Home Foreclosures Considered Separately In most of our analysis we have dealt with non-occupant owners as a whole, without distinguishing between second-home owners (owners of non-primary residences intended mainly for the owners’ occasional occupancy) and investors (owners of non-primary residences primarily meant to be rented and/or sold for gain). As discussed in the body of the paper, we have not stressed this distinction in part because key data sources such as HMDA do not make this distinction. However, the LPS data include a data field that distinguishes between the two sub-groups that make up the non-occupant category. In this appendix we examine separately the prevalence of these two types of non-occupant mortgages as well as their performance (foreclosure rate) and foreclosure impact (per housing unit). We find that mortgages to investors account for over three-fourths of the foreclosures among non-occupant mortgages, partly because they are more prevalent (accounting for about two-thirds of non-occupant mortgages), but also because they perform more poorly. However, the relative importance of second-home mortgages is above the national norm in only a few states, and Hawaii is the only state that has a foreclosure rate that is consistently higher for second homes than for investor-owned properties. Tables A3-1 to A3-4 provide information on how investors and second-home owners contributed to the overall performance, prevalence, and foreclosure impact associated with non-occupant mortgages originated in the years 2004-2007. For example, in Alabama in 2004, there were 236.35 new mortgages made to non-occupants for every 100,000 housing units, and 58.9 percent of them were originated by investors. The rate of foreclosure on Alabama’s 2004 non-occupant mortgages was 5.84 percent overall, but the performance of investor-owned mortgages was much worse, with 8.13 percent entering foreclosure by July 2009. The resulting foreclosure impact from mortgages originated in 2004 in Alabama was 13.80 foreclosures for every 100,000 housing units, with investors accounting for 82 percent of the foreclosure impact. The last two rows in each of the tables give the U.S. annual average for the year and for the entire 2004-2007 period, respectively. As shown above in Figure 11, the prevalence of mortgages to non-occupant owners rose to a peak in 2005 before subsiding in 2006 and again in 2007. However, investors’ share of non-occupant mortgages stayed near two-thirds nationally over the whole period, as shown in the second column of Tables A3-1 to A3-4. Investors’ share of non-occupant mortgage originations varies from state to state and year to year but only falls below 50 percent in a small number of states, most notably Vermont. Columns 5 to 7 of the tables show the foreclosure rates on all non-occupant mortgages as well as on mortgages to investors and second-home owners. Nationally, the foreclosure rate on mortgages to investors is significantly higher than the foreclosure rate on mortgages for second homes. The same is true in most states for all four years. This general pattern is consistent with the evidence on credit qualifications in Figure 3 (above) and the theoretical perspective on ruthlessness discussed in Appendix 1. Since investor mortgages are more prevalent, their relatively high rate of foreclosure tends to heavily influence the overall rate of foreclosure on non-occupant mortgages. Thus, the typical pattern is for the foreclosure rate on all non-occupant mortgages to be 10 to 20 percent lower than the foreclosure rate on mortgages to investors and for the foreclosure rate on second homes to be at least 25 percent lower than the overall non-occupant rate. The gap between investor and second-home foreclosure rates was widest in 2004 and narrowed somewhat thereafter. However, the performance gap between investor and second-home mortgages was narrower in a minority of states, and in a few cases (Hawaii in all four years and Arkansas, North Dakota, Wyoming, and p. 45 of 69 Utah one year each) second-home mortgages were foreclosed at a higher rate than investor mortgages. Columns 8 to 10 show the impact measure—the number of foreclosures per 100,000 housing units—for all non-occupant mortgages as well as the share of the overall impact attributed to investor versus second-home mortgages. Because mortgages to investors were more common and performed more poorly than second-home mortgages, at least 75 percent of the foreclosure impact from non-occupant mortgages nationwide can be attributed to investor mortgages. Investors’ share in the overall impact from non-occupant mortgages was especially high in a number of Midwestern and Northeastern states. Second homes played a prominent role in non-occupant foreclosures in some states, however. The leading example is Hawaii, where second-home mortgages consistently accounted for at least 55 percent of all non-occupant foreclosures. Second-home foreclosures accounted for about half of the foreclosures on non-owner-occupied properties in some years in Montana, Vermont, and Wyoming, too, and their share was above the national average in other rural states with attractive recreational amenities. Second homes also accounted for a moderately high proportion of non-occupant foreclosures in some large states that have experienced unusually high impacts from non-occupant foreclosures. For example, 30 to 40 percent of the non-occupant foreclosures in the hardest hit states, Nevada and Florida, have been on second homes, and the share in hard-hit Arizona is almost as high. Figures A3-1 to A3-8, in the format of Figure 11 above, show how the prevalence and performance of investor and second-home mortgages contributed to their respective foreclosure impacts by state. In these figures, the two types of non-occupant mortgages are benchmarked against their own 20042007 national averages. For example, in Figure A3-1, the point for Hawaii (HA) has a prevalence index (or x-axis value) of almost 2. This means that for mortgage originations in 2004, the prevalence of investor mortgages in Hawaii (per housing unit) was almost twice the 2004-2007 national average prevalence measure for investor mortgages. On the y-axis, which indexes foreclosure rates (performance), Indiana’s value of nearly 2 means that Indiana’s rate of foreclosure on investor mortgages originated in 2004 was about twice as high as the 2004-2007 national average for investor mortgages (as also shown in Table A3-1). Figures A3-5 to A3-8 are similar except that the national averages are for the prevalence and performance of second-home mortgages originated in 2004-2007. As in Figure 11, the middle of the three curves in each graph shows all points where the impact of investor (Figures A3-1 to A3-4) or second home (Figures A3-5 to A3-8) foreclosures equals the respective national average impact for 2004-2007. As before, the upper curve shows all points where the impact was three times the national average, and the lower curve shows all points where the impact was half the national average. The first four figures show that in 2005-2007, Nevada, Florida, and Arizona had the highest foreclosure impact for investor-owned mortgages. The high impact from 2005-2007 originations in these three states is a result of their above-average prevalence and high foreclosure rate on investor mortgages. States such as Georgia, Indiana, Michigan, Minnesota, Missouri, and Ohio have near to moderately above-average impacts from investor mortgages in most years that are almost entirely due to a high rate of foreclosure on these mortgages. By contrast, areas such as California, the District of Columbia, Colorado, Hawaii, Idaho, and Utah experience a similar degree of impact primarily due to an above-average prevalence of mortgages to investors. p. 46 of 69 Nevada, Florida, and Arizona are joined by Hawaii as the states with the highest relative impact from foreclosures on second homes (Figures A3-5 toA3-8). The above-average prevalence of second-home mortgages in these states was the primary cause initially, among the loans originated in 2004. Starting in 2005, the relatively poor performance of second-home mortgages in Nevada, Florida, and Arizona (and to a lesser extent Hawaii) added to their overall foreclosure impact. Several other states also have an above-average impact from second-home foreclosures in some years, but these states tend to have a high prevalence of second homes. Among these other states, Georgia is notable for an above-average rate of foreclosure on second homes, and Idaho, South Carolina, and Utah exemplify states where the prevalence of mortgages on second homes was relatively high. p. 47 of 69 Table A3-1 Foreclosure Impact Decomposition by Type of Non-Occupant Owner (NOO), 2004 (LPS data) Performance (NOO foreclosure rate) Prevalence Impact All NOO (NOO mortgages per 100000 housing units) AL AK AZ AR CA CO CT DE DC FL GA HA ID IL IN IA KS KY LA ME MD MA MI MN MS MO MT NE NV NH NJ NM NY NC ND OH OK OR PA RI SC SD TN TX UT VT VA WA WV WI WY U.S. 2004 U.S. 20042007 236.35 299.88 861.93 183.26 637.13 635.54 314.05 791.95 649.36 810.36 381.49 1216.89 727.24 249.07 281.73 134.62 223.95 171.46 184.32 400.20 498.58 385.60 293.82 318.99 137.59 305.60 435.70 155.32 1300.49 462.45 463.55 461.75 196.80 401.66 150.60 210.90 221.15 567.92 256.51 501.67 565.01 194.45 232.07 260.58 619.41 534.37 438.89 453.30 119.51 252.79 405.50 404.26 441.35 All NOO (NOO foreclosures Investor 2nd Home Investor 2nd Home per 100000 Share Share All NOO Investors 2nd-Home housing units) Share Share 58.9% 41.1% 5.84% 8.13% 2.55% 13.80 82.0% 18.0% 74.4% 25.6% 2.56% 3.11% 0.95% 7.67 90.5% 9.5% 55.9% 44.1% 2.01% 2.31% 1.64% 17.36 64.1% 35.9% 62.5% 37.5% 2.83% 4.11% 0.71% 5.19 90.6% 9.4% 75.9% 24.1% 2.41% 2.71% 1.49% 15.38 85.1% 14.9% 61.3% 38.7% 3.62% 5.17% 1.15% 22.98 87.7% 12.3% 80.0% 20.0% 3.56% 4.11% 1.35% 11.17 92.4% 7.6% 41.8% 58.2% 1.51% 3.04% 0.41% 11.97 84.1% 15.9% 83.8% 16.2% 2.27% 2.38% 1.71% 14.73 87.8% 12.2% 47.9% 52.1% 4.47% 5.66% 3.38% 36.24 60.6% 39.4% 71.5% 28.5% 9.22% 11.43% 3.66% 35.17 88.7% 11.3% 46.8% 53.2% 1.31% 0.98% 1.60% 15.95 35.1% 64.9% 63.3% 36.7% 1.90% 2.03% 1.68% 13.82 67.5% 32.5% 80.9% 19.1% 6.38% 7.38% 2.18% 15.90 93.5% 6.5% 83.1% 16.9% 16.87% 19.49% 3.98% 47.54 96.0% 4.0% 75.7% 24.3% 5.90% 7.03% 2.38% 7.94 90.2% 9.8% 86.8% 13.2% 4.75% 5.08% 2.57% 10.63 92.9% 7.1% 78.2% 21.8% 7.53% 8.66% 3.48% 12.91 89.9% 10.1% 78.3% 21.7% 4.69% 5.48% 1.83% 8.65 91.6% 8.4% 44.8% 55.2% 2.58% 3.54% 1.81% 10.34 61.4% 38.6% 71.0% 29.0% 2.27% 2.74% 1.14% 11.33 85.5% 14.5% 69.9% 30.1% 3.24% 4.22% 0.97% 12.51 91.0% 9.0% 64.0% 36.0% 13.04% 18.50% 3.31% 38.30 90.9% 9.1% 65.2% 34.8% 7.34% 10.09% 2.19% 23.41 89.6% 10.4% 63.9% 36.1% 10.12% 13.69% 3.80% 13.92 86.5% 13.5% 72.6% 27.4% 8.19% 10.57% 1.86% 25.02 93.8% 6.2% 51.0% 49.0% 1.63% 2.13% 1.11% 7.09 66.7% 33.3% 79.9% 20.1% 5.27% 6.06% 2.11% 8.18 91.9% 8.1% 60.0% 40.0% 4.74% 5.10% 4.19% 61.64 64.6% 35.4% 54.0% 46.0% 3.83% 5.42% 1.96% 17.72 76.5% 23.5% 61.6% 38.4% 3.57% 4.89% 1.45% 16.54 84.4% 15.6% 61.8% 38.2% 1.81% 1.83% 1.79% 8.37 62.3% 37.7% 65.7% 34.3% 4.97% 6.61% 1.82% 9.78 87.5% 12.5% 51.3% 48.7% 3.50% 5.44% 1.46% 14.06 79.7% 20.3% 74.4% 25.6% 2.65% 2.67% 2.59% 3.99 75.0% 25.0% 85.1% 14.9% 14.55% 16.44% 3.77% 30.68 96.1% 3.9% 78.6% 21.4% 4.97% 5.71% 2.28% 11.00 90.2% 9.8% 68.4% 31.6% 1.22% 1.36% 0.91% 6.90 76.4% 23.6% 75.6% 24.4% 4.22% 4.85% 2.28% 10.82 86.8% 13.2% 74.1% 25.9% 4.51% 5.67% 1.20% 22.63 93.1% 6.9% 39.8% 60.2% 2.85% 4.44% 1.80% 16.12 62.0% 38.0% 64.0% 36.0% 4.05% 5.63% 1.25% 7.88 88.9% 11.1% 67.5% 32.5% 7.11% 9.27% 2.61% 16.49 88.1% 11.9% 72.3% 27.7% 4.71% 5.76% 1.96% 12.27 88.5% 11.5% 62.5% 37.5% 2.72% 3.68% 1.12% 16.83 84.6% 15.4% 27.0% 73.0% 1.72% 3.64% 1.01% 9.20 57.1% 42.9% 72.6% 27.4% 1.70% 1.91% 1.15% 7.47 81.5% 18.5% 75.1% 24.9% 1.52% 1.67% 1.09% 6.90 82.2% 17.8% 51.8% 48.2% 4.34% 6.89% 1.60% 5.19 82.2% 17.8% 57.6% 42.4% 4.35% 6.33% 1.67% 11.00 83.8% 16.2% 63.9% 36.1% 1.91% 1.99% 1.76% 7.74 66.7% 33.3% 65.4% 34.6% 4.48% 5.71% 2.16% 18.10 83.3% 16.7% 66.8% 33.2% 8.50% 9.95% p. 48 of 69 5.67% 37.6 77.9% 22.1% Table A3-2 Foreclosure Impact Decomposition by Type of Non-Occupant Owner (NOO), 2005 (LPS data) Performance (NOO foreclosure rate) Prevalence Impact All NOO (NOO mortgages per 100000 housing units) AL AK AZ AR CA CO CT DE DC FL GA HA ID IL IN IA KS KY LA ME MD MA MI MN MS MO MT NE NV NH NJ NM NY NC ND OH OK OR PA RI SC SD TN TX UT VT VA WA WV WI WY U.S. 2004 U.S. 20042007 295.09 354.77 1331.47 243.16 796.37 735.63 380.02 962.47 808.80 1223.76 495.74 1544.99 1247.90 345.92 311.74 159.08 270.32 199.81 207.69 461.11 656.18 441.76 330.46 379.73 173.89 398.76 589.12 165.58 1751.33 488.29 583.62 688.03 241.94 522.63 165.55 248.96 266.92 809.00 343.37 568.32 821.15 221.60 321.28 321.90 933.12 581.44 565.89 640.07 164.43 317.01 420.26 533.16 441.35 All NOO Investor 2nd Home Share Share All NOO Investors 2nd-Home 61.3% 38.7% 7.54% 9.59% 4.29% 67.7% 32.3% 4.07% 4.36% 3.46% 62.2% 37.8% 8.47% 9.27% 7.15% 63.6% 36.4% 3.79% 4.61% 2.36% 76.7% 23.3% 8.55% 9.34% 5.96% 61.5% 38.5% 5.26% 7.10% 2.34% 80.1% 19.9% 6.90% 7.90% 2.88% 50.1% 49.9% 3.41% 5.25% 1.56% 82.9% 17.1% 4.59% 5.11% 2.07% 55.1% 44.9% 13.66% 16.10% 10.67% 68.6% 31.4% 11.60% 13.73% 6.95% 47.0% 53.0% 4.57% 3.73% 5.32% 67.9% 32.1% 3.53% 3.75% 3.06% 79.6% 20.4% 9.16% 10.37% 4.41% 83.1% 16.9% 17.67% 20.48% 3.84% 75.3% 24.7% 5.81% 7.07% 1.96% 84.8% 15.2% 4.89% 5.39% 2.04% 77.8% 22.2% 9.52% 11.44% 2.78% 76.4% 23.6% 5.86% 6.76% 2.94% 44.9% 55.1% 4.66% 5.86% 3.68% 75.7% 24.3% 4.83% 5.41% 3.01% 68.7% 31.3% 5.59% 7.33% 1.77% 66.5% 33.5% 17.91% 24.29% 5.28% 65.9% 34.1% 10.95% 15.17% 2.80% 66.5% 33.5% 10.24% 13.44% 3.89% 73.6% 26.4% 9.80% 12.10% 3.37% 47.4% 52.6% 1.15% 1.17% 1.13% 78.7% 21.3% 4.80% 5.30% 2.96% 60.6% 39.4% 14.52% 15.57% 12.90% 54.5% 45.5% 5.16% 6.96% 3.01% 67.3% 32.7% 6.45% 8.09% 3.08% 61.6% 38.4% 2.83% 2.96% 2.62% 68.1% 31.9% 7.20% 9.28% 2.77% 51.3% 48.7% 3.94% 5.31% 2.50% 73.4% 26.6% 4.17% 4.86% 2.24% 85.9% 14.1% 16.07% 17.57% 6.89% 78.1% 21.9% 6.67% 7.45% 3.88% 70.8% 29.2% 2.58% 2.73% 2.20% 79.6% 20.4% 5.13% 5.86% 2.31% 73.3% 26.7% 7.62% 8.90% 4.12% 45.7% 54.3% 5.33% 6.77% 4.11% 62.5% 37.5% 3.89% 5.19% 1.73% 67.2% 32.8% 7.49% 9.42% 3.53% 71.9% 28.1% 5.66% 6.39% 3.79% 62.0% 38.0% 3.03% 3.23% 2.71% 29.7% 70.3% 1.96% 3.01% 1.51% 75.6% 24.4% 4.44% 5.15% 2.23% 75.6% 24.4% 2.35% 2.53% 1.80% 56.3% 43.7% 8.58% 13.38% 2.39% 59.5% 40.5% 6.27% 8.76% 2.62% 61.9% 38.1% 1.01% 1.47% 0.27% 67.1% 32.9% 8.46% 9.83% 5.66% 66.8% 33.2% 8.50% 9.95% p. 49 of 69 5.67% (NOO foreclosures per 100000 housing units) 22.24 14.44 112.73 9.21 68.12 38.73 26.21 32.81 37.14 167.13 57.53 70.66 43.99 31.68 55.09 9.24 13.21 19.02 12.16 21.50 31.67 24.70 59.19 41.57 17.81 39.07 6.77 7.95 254.21 25.20 37.67 19.47 17.43 20.58 6.90 40.00 17.81 20.85 17.63 43.32 43.74 8.62 24.07 18.22 28.26 11.39 25.11 15.05 14.10 19.89 4.25 45.10 37.6 Investor 2nd Home Share Share 78.0% 22.0% 72.5% 27.5% 68.1% 31.9% 77.4% 22.6% 83.8% 16.2% 82.9% 17.1% 91.7% 8.3% 77.2% 22.8% 92.3% 7.7% 64.9% 35.1% 81.2% 18.8% 38.3% 61.7% 72.1% 27.9% 90.2% 9.8% 96.3% 3.7% 91.7% 8.3% 93.7% 6.3% 93.5% 6.5% 88.1% 11.9% 56.5% 43.5% 84.9% 15.1% 90.1% 9.9% 90.1% 9.9% 91.3% 8.7% 87.3% 12.7% 90.9% 9.1% 48.3% 51.7% 86.9% 13.1% 65.0% 35.0% 73.5% 26.5% 84.4% 15.6% 64.4% 35.6% 87.8% 12.2% 69.1% 30.9% 85.7% 14.3% 94.0% 6.0% 87.3% 12.7% 75.1% 24.9% 90.8% 9.2% 85.6% 14.4% 58.1% 41.9% 83.3% 16.7% 84.6% 15.4% 81.2% 18.8% 66.0% 34.0% 45.7% 54.3% 87.7% 12.3% 81.3% 18.8% 87.8% 12.2% 83.1% 16.9% 90.0% 10.0% 78.0% 22.0% 77.9% 22.1% Table A3-3 Foreclosure Impact Decomposition by Type of Non-Occupant Owner (NOO), 2006 (LPS data) Performance (NOO foreclosure rate) Prevalence Impact All NOO (NOO mortgages per 100000 housing units) AL AK AZ AR CA CO CT DE DC FL GA HA ID IL IN IA KS KY LA ME MD MA MI MN MS MO MT NE NV NH NJ NM NY NC ND OH OK OR PA RI SC SD TN TX UT VT VA WA WV WI WY U.S. 2004 U.S. 20042007 331.42 293.86 954.57 260.66 607.33 697.42 328.02 756.48 729.25 983.32 476.27 1188.12 1087.29 357.77 307.60 158.33 249.73 199.20 196.21 427.28 571.60 354.80 311.72 341.80 199.72 424.20 580.81 170.24 1321.68 419.51 461.49 646.46 229.29 534.50 154.34 250.14 271.47 702.29 322.52 420.17 698.37 216.79 351.83 357.83 1011.44 480.70 455.58 618.44 130.79 307.48 402.78 470.81 441.35 All NOO (NOO foreclosures Investor 2nd Home Investor 2nd Home per 100000 Share Share All NOO Investors 2nd-Home housing units) Share Share 65.0% 35.0% 8.82% 10.84% 5.07% 29.25 79.9% 20.1% 69.9% 30.1% 6.32% 7.13% 4.44% 18.57 78.8% 21.2% 63.0% 37.0% 14.51% 16.53% 11.07% 138.53 71.8% 28.2% 66.0% 34.0% 6.00% 7.49% 3.10% 15.63 82.4% 17.6% 76.7% 23.3% 14.15% 15.17% 10.79% 85.93 82.2% 17.8% 63.7% 36.3% 6.34% 8.24% 3.00% 44.22 82.8% 17.2% 79.1% 20.9% 11.60% 13.65% 3.87% 38.05 93.0% 7.0% 51.9% 48.1% 6.28% 9.32% 3.01% 47.54 76.9% 23.1% 83.4% 16.6% 8.39% 9.24% 4.09% 61.15 91.9% 8.1% 55.4% 44.6% 24.70% 28.50% 19.99% 242.86 63.9% 36.1% 66.4% 33.6% 14.15% 16.69% 9.13% 67.37 78.3% 21.7% 48.4% 51.6% 8.00% 7.33% 8.62% 94.99 44.4% 55.6% 70.9% 29.1% 7.56% 8.01% 6.46% 82.20 75.1% 24.9% 80.5% 19.5% 12.43% 14.10% 5.54% 44.47 91.3% 8.7% 83.0% 17.0% 19.34% 22.34% 4.72% 59.50 95.9% 4.1% 70.8% 29.2% 5.97% 7.07% 3.30% 9.45 83.9% 16.1% 82.8% 17.2% 5.84% 6.45% 2.89% 14.58 91.5% 8.5% 78.5% 21.5% 8.77% 10.16% 3.71% 17.47 90.9% 9.1% 76.6% 23.4% 7.44% 8.69% 3.36% 14.61 89.4% 10.6% 42.5% 57.5% 5.11% 8.69% 2.47% 21.85 72.2% 27.8% 79.1% 20.9% 9.29% 10.44% 4.95% 53.09 88.9% 11.1% 67.4% 32.6% 7.60% 9.69% 3.28% 26.95 85.9% 14.1% 71.0% 29.0% 21.37% 27.61% 6.09% 66.63 91.7% 8.3% 65.7% 34.3% 14.81% 19.80% 5.27% 50.63 87.8% 12.2% 69.1% 30.9% 11.03% 12.84% 6.99% 22.04 80.4% 19.6% 73.7% 26.3% 10.12% 12.33% 3.94% 42.95 89.8% 10.2% 50.4% 49.6% 3.23% 3.48% 2.97% 18.75 54.3% 45.7% 79.0% 21.0% 5.91% 7.01% 1.81% 10.07 93.6% 6.4% 62.2% 37.8% 21.42% 22.04% 20.40% 283.09 64.0% 36.0% 50.4% 49.6% 6.99% 10.18% 3.75% 29.34 73.4% 26.6% 68.6% 31.4% 10.66% 13.14% 5.22% 49.18 84.6% 15.4% 59.4% 40.6% 4.84% 5.13% 4.43% 31.32 62.9% 37.1% 67.2% 32.8% 9.85% 12.87% 3.68% 22.59 87.7% 12.3% 54.1% 45.9% 5.23% 6.62% 3.59% 27.94 68.4% 31.6% 71.8% 28.2% 2.32% 2.93% 0.75% 3.57 90.9% 9.1% 87.6% 12.4% 18.48% 20.28% 5.76% 46.23 96.1% 3.9% 79.9% 20.1% 8.23% 9.29% 3.99% 22.33 90.3% 9.7% 72.6% 27.4% 5.64% 5.89% 4.98% 39.62 75.8% 24.2% 79.8% 20.2% 7.57% 8.55% 3.69% 24.41 90.2% 9.8% 70.9% 29.1% 13.55% 17.55% 3.82% 56.94 91.8% 8.2% 46.6% 53.4% 7.25% 9.64% 5.16% 50.61 61.9% 38.1% 61.3% 38.7% 4.71% 6.82% 1.35% 10.20 88.9% 11.1% 66.5% 33.5% 9.23% 11.51% 4.71% 32.47 82.9% 17.1% 70.4% 29.6% 7.20% 7.99% 5.31% 25.76 78.2% 21.8% 65.1% 34.9% 5.69% 5.31% 6.39% 57.50 60.8% 39.2% 33.0% 67.0% 3.76% 6.72% 2.31% 18.09 58.9% 41.1% 75.5% 24.5% 7.76% 9.03% 3.85% 35.35 87.8% 12.2% 75.4% 24.6% 5.22% 5.49% 4.42% 32.30 79.2% 20.8% 55.7% 44.3% 7.75% 10.31% 4.53% 10.14 74.2% 25.8% 59.2% 40.8% 8.55% 11.64% 4.06% 26.29 80.6% 19.4% 63.4% 36.6% 2.49% 2.95% 1.70% 10.04 75.0% 25.0% 67.8% 32.2% 12.66% 14.4% 9.1% 59.59 76.9% 23.1% 66.8% 33.2% 8.50% 9.95% p. 50 of 69 5.67% 37.6 77.9% 22.1% Table A3-4 Foreclosure Impact Decomposition by Type of Non-Occupant Owner (NOO), 2007 (LPS data) Performance (NOO foreclosure rate) Prevalence Impact All NOO (NOO mortgages per 100000 housing units) AL AK AZ AR CA CO CT DE DC FL GA HA ID IL IN IA KS KY LA ME MD MA MI MN MS MO MT NE NV NH NJ NM NY NC ND OH OK OR PA RI SC SD TN TX UT VT VA WA WV WI WY U.S. 2004 277.40 258.30 650.07 217.81 471.98 625.62 260.07 638.17 574.90 567.10 369.77 891.39 765.72 285.06 218.22 150.36 216.26 164.79 169.14 387.44 437.71 314.95 238.60 262.81 186.87 359.18 517.27 142.67 757.73 358.72 356.79 493.10 189.32 440.89 125.26 190.97 226.87 559.25 248.69 328.02 500.71 209.65 295.85 293.76 820.55 412.93 359.28 538.58 107.86 256.51 363.03 358.45 U.S. 20042007 441.35 All NOO Investor 2nd Home Share Share All NOO Investors 2nd-Home 63.2% 36.8% 5.25% 6.54% 3.03% 65.4% 34.6% 2.74% 2.31% 3.57% 61.9% 38.1% 11.60% 14.42% 7.02% 65.3% 34.7% 3.78% 5.14% 1.23% 75.1% 24.9% 7.99% 9.00% 4.96% 64.6% 35.4% 3.59% 4.43% 2.06% 76.5% 23.5% 5.56% 6.43% 2.73% 48.0% 52.0% 2.94% 4.29% 1.71% 81.0% 19.0% 5.02% 5.89% 1.29% 50.5% 49.5% 17.93% 22.42% 13.35% 67.1% 32.9% 8.78% 9.74% 6.82% 47.6% 52.4% 4.41% 4.00% 4.77% 67.8% 32.2% 6.27% 7.08% 4.57% 79.8% 20.2% 7.49% 8.47% 3.60% 80.5% 19.5% 8.58% 9.92% 3.04% 73.1% 26.9% 2.87% 3.37% 1.50% 84.6% 15.4% 3.30% 3.50% 2.21% 76.2% 23.8% 4.58% 5.18% 2.68% 77.3% 22.7% 4.86% 5.51% 2.66% 34.9% 65.1% 3.56% 7.01% 1.71% 76.1% 23.9% 7.06% 8.08% 3.83% 64.7% 35.3% 4.05% 5.34% 1.68% 67.2% 32.8% 14.36% 19.36% 4.12% 63.1% 36.9% 9.88% 13.71% 3.34% 72.7% 27.3% 7.63% 8.44% 5.49% 72.6% 27.4% 6.66% 8.23% 2.49% 49.6% 50.4% 2.26% 2.42% 2.12% 72.0% 28.0% 2.33% 2.74% 1.28% 60.3% 39.7% 14.01% 15.67% 11.49% 45.8% 54.2% 5.07% 7.38% 3.11% 61.5% 38.5% 7.35% 9.63% 3.72% 60.6% 39.4% 4.32% 4.69% 3.75% 63.5% 36.5% 5.23% 6.99% 2.15% 55.4% 44.6% 3.14% 3.59% 2.58% 64.3% 35.7% 0.51% 0.40% 0.72% 86.6% 13.4% 8.95% 9.80% 3.47% 76.4% 23.6% 3.45% 3.63% 2.88% 70.9% 29.1% 4.58% 5.17% 3.17% 77.5% 22.5% 3.86% 4.34% 2.22% 63.9% 36.1% 6.36% 9.10% 1.50% 43.3% 56.7% 5.12% 6.93% 3.73% 62.3% 37.7% 1.74% 2.14% 1.06% 66.9% 33.1% 5.50% 5.69% 5.10% 72.3% 27.7% 3.32% 3.53% 2.77% 69.3% 30.7% 6.69% 7.03% 5.93% 37.6% 62.4% 2.72% 3.11% 2.49% 73.3% 26.7% 4.50% 5.21% 2.55% 76.0% 24.0% 4.14% 4.48% 3.09% 51.2% 48.8% 5.25% 8.42% 1.94% 57.2% 42.8% 6.25% 8.28% 3.54% 60.6% 39.4% 0.91% 0.56% 1.44% 66.6% 33.4% 7.65% 8.79% 5.36% 66.8% 33.2% 8.50% 9.95% p. 51 of 69 5.67% (NOO foreclosures per 100000 housing units) 14.55 7.09 75.38 8.23 37.72 22.47 14.46 18.78 28.85 101.67 32.46 39.27 48.02 21.35 18.72 4.31 7.14 7.55 8.23 13.78 30.92 12.77 34.27 25.97 14.26 23.91 11.71 3.33 106.13 18.18 26.23 21.29 9.89 13.84 0.64 17.10 7.83 25.63 9.60 20.85 25.62 3.64 16.26 9.76 54.91 11.24 16.16 22.32 5.66 16.04 3.30 27.41 37.6 Investor 2nd Home Share Share 78.8% 21.2% 55.0% 45.0% 76.9% 23.1% 88.7% 11.3% 84.5% 15.5% 79.7% 20.3% 88.5% 11.5% 69.9% 30.1% 95.1% 4.9% 63.1% 36.9% 74.4% 25.6% 43.2% 56.8% 76.6% 23.4% 90.3% 9.7% 93.1% 6.9% 86.0% 14.0% 89.7% 10.3% 86.1% 13.9% 87.6% 12.4% 68.8% 31.3% 87.0% 13.0% 85.3% 14.7% 90.6% 9.4% 87.5% 12.5% 80.3% 19.7% 89.7% 10.3% 52.9% 47.1% 84.6% 15.4% 67.4% 32.6% 66.7% 33.3% 80.5% 19.5% 65.8% 34.2% 85.0% 15.0% 63.4% 36.6% 50.0% 50.0% 94.8% 5.2% 80.3% 19.7% 79.9% 20.1% 87.1% 12.9% 91.5% 8.5% 58.7% 41.3% 76.9% 23.1% 69.3% 30.7% 76.8% 23.2% 72.8% 27.2% 42.9% 57.1% 84.9% 15.1% 82.1% 17.9% 82.0% 18.0% 75.8% 24.2% 37.5% 62.5% 76.5% 23.5% 77.9% 22.1% Figure A3-1 Relative Investor Foreclosure Impact per Housing Unit 2 3 2004 Investor Purchase and Refinance Originations (LPS) IN MI OH MS 0 1 GA MO MN TN KY AL IL WVIA NY WI TX NE SD RIFL LAOK NC NHNJ CO KS PA MA ARVTME SCCT UT AK DE MD ND MTWY NMVA WA OR 0 NV CA IDAZ DC HA 1 2 3 Relative Investor Mortgage Prevalence (2004-07 Nat'l Avg.=1) 4 Figure A3-2 Relative Investor Foreclosure Impact per Housing Unit 3 2005 Investor Purchase and Refinance Originations (LPS) MI 2 IN OH FL MN MS WV 1 KY NV GA MO IL RI NJ SC CO MD VA DE NM WA CA AZ DC UT OR HA ID 0 ALTN NY WI CT IA LA OK NHMA TX PA ME KSNC NE SD ND AR AK VT WY MT 0 1 2 3 Relative Investor Mortgage Prevalence (2004-07 Nat'l Avg.=1) p. 52 of 69 4 Figure A3-3 Relative Investor Foreclosure Impact per Housing Unit 3 2006 Investor Purchase and Refinance Originations (LPS) FL MI IN NV 2 OH MN RI GA AZ CA 1 CTIL NJ NY MS MO WI TN AL MD WV KY NH MA SC DE DC VA LAMEOKPA CO ARAK TX HA IASD NE VT KS NC OR UT NM WA ID WYMT 0 ND 0 1 2 3 Relative Investor Mortgage Prevalence (2004-07 Nat'l Avg.=1) 4 Figure A3-4 Relative Investor Foreclosure Impact per Housing Unit 3 2007 Investor Purchase and Refinance Originations (LPS) 2 FL MI AZ 1 MN WV IN NJGA OH RI CA IL MO MD MS WI ID UT MENH NY SC ALCT DC LA MA VA KY OR AR TN NM WA CO PA DE HA KSTXNC IA OK NE VT MT SDAK 0 ND 0 NV WY 1 2 3 Relative Investor Mortgage Prevalence (2004-07 Nat'l Avg.=1) p. 53 of 69 4 Figure A3-5 Relative 2nd-Home Foreclosure Impact per Housing Unit 1 2 3 4 2004 2nd-Home Purchase and Refinance Originations (LPS) 0 IN MS GA OH KY MI ND TNAL KS IAOK MN ILPA NE TX NH MO LA NY ME ID DC WI WY WV CANM NJNC CT SD RIMD VA CO UT MA AK WA OR MT AR 0 NV FL SC AZ VT HA DE 2 4 Relative 2nd-Home Mortgage Prevalence (2004-07 Nat'l Avg.=1) 6 Figure A3-6 Relative 2nd-Home Foreclosure Impact per Housing Unit 3 4 2005 2nd-Home Purchase and Refinance Originations (LPS) 2 NV FL 1 OH AZ GA CA 0 MI IL AL RI SC MS TX OK IN TN AK MDNJ NH ME MO ID NE LA CT KY MN NY UT WI NM NC WV NDPAAR VA OR CO KS IA WA MA SD DC DE VT MT WY 0 2 4 Relative 2nd-Home Mortgage Prevalence (2004-07 Nat'l Avg.=1) p. 54 of 69 HA 6 Figure A3-7 Relative 2nd-Home Foreclosure Impact per Housing Unit 4 2006 2nd-Home Purchase and Refinance Originations (LPS) NV 2 3 FL AZ CA GA HA 0 1 MS ID UT OH IL MITX MN NJ OR SC AL MD IN TN WVAK DC NM WA WI OK MO CT VA RI NH KY NY MA NC LA IAPA AR CO MT DE KS ME VT NE WY SD ND 0 2 4 Relative 2nd-Home Mortgage Prevalence (2004-07 Nat'l Avg.=1) 6 Figure A3-8 Relative 2nd-Home Foreclosure Impact per Housing Unit 3 4 2007 2nd-Home Purchase and Refinance Originations (LPS) FL 2 NV 1 GA AZ UT ID 0 MS TN CA MI MD NJ SC IL AK WIWAOR NM OH MN NH INOK AL CTTX KY LA NC VA MO VT PA KSWV NY MT COME DE MA IA AR RI WY NE DC ND SD 0 HA 2 4 Relative 2nd-Home Mortgage Prevalence (2004-07 Nat'l Avg.=1) p. 55 of 69 6 Appendix 4: The Prevalence, Performance and Impact (per Housing Unit) of Non-Occupant Mortgages Originated in 2004, 2005, and 2007 The results here are for mortgages originated in 2004, 2005, and 2007. The maps have the same format as Figures 12-14, which was explained in the body of the paper. Table A4-1 summarizes the data for the maps in this appendix as well as for Figures 11-14. Table A4-1 Indices of the Relative Performance, Prevalence, and Impact of Non-Owner-Occupied (NOO) Mortgages and Foreclosures, by State and Year Mortgage Was Originated State AL AK AZ AR CA CO CT DE DC FL GA HA ID IL IN IA KS KY LA ME MD MA MI MN MS MO MT NE NV NH NJ NM NY NC ND OH OK OR PA RI SC SD TN TX UT VT VA WA WV WI WY Performance Index (Foreclosures per mortgage, for NOO mortgages, by year of origination, as a percentage of the 2004-07 national average) 2004 2005 2006 2007 88 103 62 68 30 48 74 32 24 99 170 136 33 44 70 44 28 100 166 94 42 62 74 42 42 81 136 65 18 40 74 35 27 54 98 59 52 160 290 210 108 136 166 103 15 54 94 52 22 41 89 74 75 107 146 88 198 207 227 101 69 68 70 34 56 57 68 39 88 112 103 54 55 69 87 57 30 55 60 42 27 57 109 83 38 66 89 48 153 210 251 168 86 128 174 116 119 120 129 90 96 115 119 78 19 13 38 27 62 56 69 27 56 170 251 164 45 61 82 59 42 76 125 86 21 33 57 51 58 84 116 61 41 46 61 37 31 49 27 6 171 188 217 105 58 78 96 40 14 30 66 54 49 60 89 45 53 89 159 75 33 62 85 60 48 46 55 20 83 88 108 64 55 66 84 39 32 36 67 78 20 23 44 32 20 52 91 53 18 28 61 49 51 101 91 62 51 74 100 73 22 12 29 11 Prevalence Index (Mortgages per housing unit, for NOO mortgages, by year of origination, as a percentage of the 2004-07 national average) 2004 2005 2006 2007 67 75 63 54 68 80 67 59 195 302 216 147 42 55 59 49 144 180 138 107 144 167 158 142 71 86 74 59 179 218 171 145 147 183 165 130 184 277 223 128 86 112 108 84 276 350 269 202 165 283 246 173 56 78 81 65 64 71 70 49 31 36 36 34 51 61 57 49 39 45 45 37 42 47 44 38 91 104 97 88 113 149 130 99 87 100 80 71 67 75 71 54 72 86 77 60 31 39 45 42 69 90 96 81 99 133 132 117 35 38 39 32 295 397 299 172 105 111 95 81 105 132 105 81 105 156 146 112 45 55 52 43 91 118 121 100 34 38 35 28 48 56 57 43 50 60 62 51 129 183 159 127 58 78 73 56 114 129 95 74 128 186 158 113 44 50 49 48 53 73 80 67 59 73 81 67 140 211 229 186 121 132 109 94 99 128 103 81 103 145 140 122 27 37 30 24 57 72 70 58 92 95 91 82 p. 56 of 69 Impact Index (NOO foreclosures per housing unit, by year of origination, as a percentage of the 2004-07 national average) 2004 2005 2006 2007 59 78 39 37 20 38 49 19 46 300 368 200 14 24 42 22 41 181 228 100 61 103 118 60 30 70 101 38 32 87 126 50 39 99 162 77 96 444 645 270 93 153 179 86 42 188 252 104 37 117 218 128 42 84 118 57 126 146 158 50 21 25 25 11 28 35 39 19 34 51 46 20 23 32 39 22 27 57 58 37 30 84 141 82 33 66 72 34 102 157 177 91 62 110 135 69 37 47 59 38 66 104 114 64 19 18 50 31 22 21 27 9 164 675 752 282 47 67 78 48 44 100 131 70 22 52 83 57 26 46 60 26 37 55 74 37 11 18 9 2 82 106 123 45 29 47 59 21 18 55 105 68 29 47 65 26 60 115 151 55 43 116 134 68 21 23 27 10 44 64 86 43 33 48 68 26 45 75 153 146 24 30 48 30 20 67 94 43 18 40 86 59 14 37 27 15 29 53 70 43 21 11 27 9 Figure A4-1: Non-Owner-Occupied Mortgage Prevalence per Housing Unit, Relative to 2004-2007 U.S. Average (LPS data for mortgages originated in 2004) Figure A4-2: Non-Owner-Occupied Mortgage Foreclosure Rates, Relative to 2004-2007 U.S. Average (LPS data for mortgages originated in 2004) p. 57 of 69 Figure A4-3: Non-Owner-Occupied Mortgage Foreclosure Impact per Housing Unit, Relative to U.S. 2004-2007 Average (LPS data for mortgages originated in 2004) Figure A4-4: Non-Owner-Occupied Mortgage Prevalence per Housing Unit, Relative to 2004-2007 U.S. Average (LPS data for mortgages originated in 2005) p. 58 of 69 Figure A4-5: Non-Owner-Occupied Mortgage Foreclosure Rates, Relative to 2004-2007 U.S. Average (LPS data for mortgages originated in 2005) Figure A4-6: Non-Owner-Occupied Mortgage Foreclosure Impact per Housing Unit, Relative to U.S. 2004-2007 Average (LPS data for mortgages originated in 2005) p. 59 of 69 Figure A4-7: Non-Owner-Occupied Mortgage Prevalence per Housing Unit, Relative to 2004-2007 U.S. Average (LPS data for mortgages originated in 2007) Figure A4-8: Non-Owner-Occupied Mortgage Foreclosure Rates, Relative to 2004-2007 U.S. Average (LPS data for mortgages originated in 2007) p. 60 of 69 Figure A4-9: Non-Owner-Occupied Mortgage Foreclosure Impact per Housing Unit, Relative to U.S. 2004-2007 Average (LPS data for mortgages originated in 2007) p. 61 of 69 Appendix 5: Non-Occupant Mortgage Prevalence and Foreclosure Impact Measured per Mortgage Originated Instead of Per Housing Unit, 2004-2006 This paper presents measures of non-occupant mortgage prevalence and non-occupant foreclosure impact based on a state’s volume of non-occupant mortgage originations and non-occupant foreclosures relative to the total number of housing units in the state. An alternative measure is to express these volumes relative to the total volume of mortgage originations.50 These alternative measurements are presented here for mortgages originated in 2004-2007, using formats similar to those of Figures 11, 12, and 14 but with total mortgage originations in the denominator instead of total housing units.51 Compared to the per housing unit measures in Figures 11, 12, and 14, the per mortgage measures presented here tend to show a higher impact in Midwestern states like Ohio and Indiana, where mortgage originations per housing unit in 2004-2007 were lower than in Western states like Arizona and Idaho. The fact that 2004-2007 mortgage originations per housing unit were higher in Arizona, than in Indiana and Ohio may be related to the faster rate of growth in population and new housing construction in recent years in Arizona. 2004 Figure A5-1 Relative Non-Occupant Foreclosure Impact per Mortgage 2 3 2004 Home Purchase and Refinance Originations (LPS) IN OH MI HA 0 1 MS GA MO KY MNTN IL IA AL NE NY OK KS NV TX LA RI FL WI WV PA SD NJ NH CO CT NC MA SC UT NDAR ME AK CA MD DC NM AZ WY ID VT VA MT WA DE OR 0 1 2 3 Relative Non-Occupant Mortgage Prevalence (U.S.=1) 4 Note: Middle line represents an impact equal to the LPS data 2004-07 national average of 8.5 non-occupant foreclosures per 1000 home mortgages (owner-occupied and non-owner occupied) . The lower and upper lines represent impacts of, respectively, half and three times that national average impact. 50 In all cases, the mortgages considered are first-lien housing purchase and refinance mortgages on single-family housing units. 51 Figure 13 is not affected by the choice of total mortgages as the denominator for defining prevalence and impact. p. 62 of 69 Figure A5-2 Figure A5-3 p. 63 of 69 2005 Figure A5-4 Relative Non-Occupant Foreclosure Impact per Mortgage 3 2005 Home Purchase and Refinance Originations (LPS) 2 MI IN OH NV FL SC HA 0 1 GA MN MS MO IL KY WV CA AZ RI AL TN CTNY OK NJ WI LATX IA MA NH PA CO ME NE KS MD DC VA ND NC SDAK AR ID DE UT NM OR WA VT MT WY 0 1 2 3 Relative Non-Occupant Mortgage Prevalence (U.S.=1) Note: Middle line represents an impact equal to the LPS data 2004-07 national average of 8.5 non-occupant foreclosures per 1000 home mortgages (owner-occupied and non-owner occupied) . The lower and upper lines represent impacts of, respectively, half and three times that national average impact. p. 64 of 69 4 Figure A5-5 Figure A5-6 p. 65 of 69 2006 Figure A5-7 Relative Non-Occupant Foreclosure Impact per Mortgage 3 2006 Home Purchase and Refinance Originations (LPS) FL NV MI 2 IN OH 0 1 MN CA GA AZ RI IL CT NJMS NY MO MD AL DC KYWITN OK VA WV MA PA TX ID SCHA LA NH AK AR CO DE IA NEKS UT OR WA NC ME NM SD VT MT WY ND 0 1 2 3 Relative Non-Occupant Mortgage Prevalence (U.S.=1) Note: Middle line represents an impact equal to the LPS data 2004-07 national average of 8.5 non-occupant foreclosures per 1000 home mortgages (owner-occupied and non-owner occupied) . The lower and upper lines represent impacts of, respectively, half and three times that national average impact. p. 66 of 69 4 Figure A5-8 Figure A5-9 p. 67 of 69 2007 Figure A5-10 Relative Non-Occupant Foreclosure Impact per Mortgage 3 2007 Home Purchase and Refinance Originations (LPS) 2 FL MI NV AZ 0 1 MN IN OHGA CA MS IL NJ MDMO UT RI WI ID CT NY TN WV AL SC NH DC LA KY VAWA ORNM HA MA PA AR TX CO ME OK KS NC DE VT IA NEAK MT SD WY ND 0 1 2 3 Relative Non-Occupant Mortgage Prevalence (U.S.=1) Note: Middle line represents an impact equal to the LPS data 2004-07 national average of 8.5 non-occupant foreclosures per 1000 home mortgages (owner-occupied and non-owner occupied) . The lower and upper lines represent impacts of, respectively, half and three times that national average impact. p. 68 of 69 4 Figure A5-11 Figure A5-12 p. 69 of 69