Document Sample

Deficiency Judgments and Borrower Maintenance: Theory and Evidence John Harding,* Thomas J. Miceli,** and C.F. Sirmans* Journal of Housing Economics, 9(4), December 2000, 267-286 Abstract It is well known when property rights to an asset are divided, individual rightholders may have adequate incentives to invest in proper maintenance. In this paper, we examine how mortgage laws affect the nature of the lender’s claim to the house, and how that claim in turn affects the incentives of borrowers to invest in home maintenance. The specific law that we examine concerns the right of lenders to pursue a borrower’s non-housing wealth in the event of default if the value of the house is less than the mortgage balance. Most states allow lenders to collect such “deficiency judgments,” while others either prohibit, or make it difficult to obtain them. The theoretical model developed in this paper predicts that borrowers will maintain at a higher rate when lenders are allowed to seek deficiency judgments. Intuitively, when borrowers’ non-housing wealth is at risk, whey have an incentive to invest more in maintenance in order to reduce the likelihood that the value of the property will fall below the mortgage balance. We attempt to measure this effect using data on household maintenance obtained from the American Housing Survey along with information on differences in mortgage laws across states. We estimate a three equation simultaneous system relating maintenance expenditures, house value, and mortgage rates. The results provide confirmation that variation in mortgage laws affect homeowner maintenance in the manner predicted by the theory. January 2000 *Department of Finance and Real Estate Center School of Business Administration University of Connecticut Storrs, CT 06269 **Department of Economics University of Connecticut Storrs, CT 06269 1 Deficiency Judgments and Borrower Maintenance: Theory and Evidence I. Introduction It is well known that when property rights to an asset are divided, individual rightholders may have inadequate incentives to invest in proper maintenance. This is true when ownership is divided at a point in time (as with joint ownership), or over time (as in a leasing arrangement). It is also true when the asset is used as collateral for a loan--as when a homebuyer purchases a house with a mortgage--for in that case, the lender has a claim on the asset in the event of default. In this paper, we examine how mortgage laws affect the nature of the lender’s claim to the house, and how that claim in turn affects the incentives of borrowers to invest in home maintenance. The specific law that we examine concerns the right of lenders to pursue a borrower’s non-housing wealth in the event of default if the value of the house is less than the mortgage balance. Most states allow lenders to collect such “deficiency judgments,” while others either prohibit, or make it difficult to obtain them. The theoretical model developed in this paper predicts that borrowers will maintain at a higher rate when lenders are allowed to seek deficiency judgments. Intuitively, when borrowers’ non-housing wealth is at risk, they have an incentive to invest more in maintenance in order to reduce the likelihood that the value of the property will fall below the mortgage balance. It should be noted that the reduced incentive to invest when deficiency judgments are prohibited is partially offset by the law of waste, which allows lenders to seek compensation for abuse of the mortgaged property. However, courts traditionally have been unwilling to uphold a claim of waste for mere failure to maintain (what Stein (1998: p. 1211) calls “passive waste”), and further, courts tend to be skeptical of lenders’ efforts to use the law of waste to circumvent 2 anti-deficiency judgment statutes (Burke, 1989: pp. 394-395).1 Ultimately, therefore, we expect the availability of deficiency judgments to have the predicted positive effect on maintenance. We attempt to measure this effect using data on household maintenance obtained from the American Housing Survey along with information on differences in mortgage laws across states. We estimate a three equation simultaneous system relating maintenance expenditures, house value, and mortgage rates. The results provide confirmation that variation in mortgage laws affect homeowner maintenance in the manner predicted by the theory. The remainder of the paper is organized as follows. Section II sets up the model and derives the first-best level of maintenance, which serves as a benchmark. Section III then examines the impact of mortgage laws by comparing the maintenance level of borrowers when deficiency judgments are allowed and when they are prohibited. Section IV provides an empirical test of the predictions of the model. Finally, Section V offers concluding remarks. II. The Model and First-Best Maintenance As a benchmark, we begin by deriving the optimal maintenance level of an owner- occupier who does not have a mortgage. Since there is no risk of default, the owner will internalize the full costs and benefits of his maintenance. Thus, the resulting level of maintenance is first-best. We consider a risk-neutral homeowner who maximizes a two-period utility function of the form2 EU = x + v(h(m)) + δE ( w) , (1) 1 See Stein (1998) for a discussion of those situations in which lenders have been allowed by courts to recover against a borrower’s assets under the law of waste, despite anti-deficiency judgment provisions in the loan. His analysis, however, is primarily in the context of commercial loans. 2 See, for example, Henderson and Ioannides (1983). Since our focus is on the maintenance decisions of borrowers, we ignore risk-sharing aspects of the mortgage contract. 3 where x is non-housing consumption in period one, v(h(m)) is the value of housing services as a function of housing quality (v′>0, v″<0), w is second period wealth, and δ is the discount rate. Housing quality depends on maintenance expenditures, m, (h′>0, h″<0), which we view as a one-time investment, made in period one. The first period budget constraint is given by y1 = x + m +P0, where y1 is first period income (wealth), and P0 is the purchase price of the house (a sunk cost). Second period wealth is defined to be w=y2+Ph(m), where y2 is second period income (non-housing wealth), and P is the second period market value of a unit of housing. We assume that y2 is known but that P is a random variable from the perspective of period one, which accounts for the expectations operator in (1). Note that period one maintenance expenditures affect both the value of housing for consumption in period one, and the market value of housing as a store of wealth in period two. Substituting the budget constraint and definition of wealth into (1) yields EU = y1 − m − P0 + v(h(m)) + δE[ y 2 + Ph(m)] . (2) The owner’s problem is to choose maintenance to maximize (2). The resulting first-order condition for optimal maintenance, m*, is [v ′ + δE ( P )]h ′ = 1 . (3) This condition says that the homeowner should equate the marginal benefit of maintenance— consisting of the marginal consumption benefit plus the marginal benefit on housing wealth—to the marginal cost ($1). Equation (3) will serve as the benchmark for the analysis of borrower incentives under different mortgage laws. 4 III. The Impact of Mortgage Laws on Borrower Maintenance Suppose now that the homeowner/borrower purchases the house at the beginning of period one with a mortgage in the amount L3 and must repay L(1+r) in period two, where r is the mortgage interest rate. We assume that y2<L(1+r) so that the borrower must sell (or refinance) the house in order to pay off the mortgage. Depending on the realized value of P, three outcomes are possible: (1) Ph(m)≥L(1+r): the proceeds from the sale are sufficient to cover the mortgage; (2) y2+Ph(m)≥L(1+r)>Ph(m): the proceeds from the sale are insufficient to cover the mortgage, but the proceeds plus the borrower’s non-housing wealth are sufficient; and (3) y2+Ph(m)<L(1+r): the proceeds from the sale plus the borrower’s non-housing wealth are insufficient to cover the mortgage. Our interest is in what happens in cases two and three. Specifically, does the lender have the right to seek a deficiency judgment against the borrower’s non-housing wealth in the event that the sale price fails to cover the mortgage balance? So-called recourse mortgages permit deficiency judgments, whereas non-recourse mortgages do not (Gibson, Karp, and Klayman, 1992: Ch. 18). In the next two sections we examine the period-one maintenance incentives of borrowers under both types of mortgages. A. Deficiency Judgments Allowed Forty-two states in the U.S. allow deficiency judgments, meaning that lenders can seek to cover any deficit at foreclosure out of the borrower’s non-housing wealth, y2 (State Legislative Topics, 1992). Thus, when Ph(m)<L(1+r), borrowers are liable up to the full amount of their period two resources of y2+Ph(m). As a result, their period two wealth is max[ y 2 + Ph(m) − L(1 + r ),0] , while lenders receive revenue of max[ L(1 + r ), y 2 + Ph(m)] . 3 We assume for simplicity that the homeowner finances 100% of the purchase price. 5 Figure 1 graphs borrower wealth (bold solid line) and lender revenue (bold dashed line) as functions of the sale price, Ph(m) (given m). Note that the horizontal segment of the borrower’s wealth at zero reflects the fact that the latter’s liability is limited to his second period wealth (including non-housing wealth) in the event of a deficit. This is the key with respect to borrower maintenance incentives. The borrower’s expected utility when deficiency judgments are allowed is given by ∞ EU D = y1 − m + v (h( m)) + δ ò [ y 2 + Ph( m) − L(1 + r )]dF ( P ) , (4) A where F(P) is the distribution function of P, and A ≡ [ L(1 + r ) − y 2 ] / h(m) . The integral term represents expected wealth in period two, where the lower bound of A>0 reflects the limited liability of borrowers in the event of a deficit.4 Note, however, that the right of lenders to seek deficiency judgments against y2 in effect reduces this lower bound (i.e., ∂A/∂y2<0). This turns out to have an important effect on borrower incentives. Optimal first period maintenance is found by differentiating (4) with respect to m to obtain ∞ [v ′ + δ ò PdF ( P)]h ′ = 1 . (5) A Let mD denote the solution to this condition. Comparing (5) and (3) shows that mD<m*, given A>0.5 Further, maintenance is decreasing the greater is the range of limited liability. The reason is that the borrower fully internalizes the marginal benefit of maintenance over the range where P>A because his wealth is increasing dollar-for-dollar in the value of the house over this range. In contrast, his wealth is zero, and hence independent of m, over the range where P<A. Thus, it is 4 Thus, the kink in the borrower’s wealth in Figure 1 corresponds to A. That is, at the kink Ph(m)=L(1+r)-y2. 5 Note, in particular, that if A=0, (5) reduces to (3). 6 the borrower’s limited liability in the default state that leads to undermaintenance. And, as noted above, the fact that the lender has recourse to the borrower’s non-housing assets reduces the range of limited liability and hence results in a higher level of maintenance. B. Deficiency Judgments Not Allowed Eight states prohibit, or at least make it difficult for lenders to seek, deficiency judgments against borrowers in the event of a deficit at foreclosure.6 The lender’s recovery is therefore limited to Ph(m) even when this amount falls short of L(1+r). (We assume for now that lenders charge the same rate of interest regardless of whether or not they can seek deficiency judgments.) As a result, the borrower’s period two wealth is y2+ max[ Ph(m) − L(1 + r ),0] , and the lender’s revenue is max[ L(1 + r ), Ph(m)] . Figure 2 graphs these values, again as a function of Ph(m). Note that the range of limited liability (the horizontal segment of the borrower’s wealth at y2) is larger here compared to Figure 1 due to the prohibition on deficiency judgments, given m. Thus, for any level of maintenance, the range of P over which there is a deficit (which in this model is equivalent to default) is larger here than when deficiency judgments are allowed.7 Proceeding as above, we can write the borrower’s expected utility in this case as ∞ EU N = y1 − m + v(h(m)) + δ { y 2 + ò [Ph(m) − L(1 + r )]dF ( P )} , (6) B where B ≡ L(1 + r ) / h( m) . The first order condition for optimal maintenance in this case, denoted mN, is given by ∞ [v ′ + δ ò PdF ( P)]h ′ = 1 . (7) B 6 The eight states are Alaska, Arizona, California, Minnesota, Montana, North Dakota, Oregon, and Washington. Alaska, California, and Minnesota allow deficiency judgments when the lender uses judicial proceedings against the borrower, but this rarely occurs according to the Mortgage Bankers Association (State Legislative Topics, 1992). 7 Jones (1993) finds evidence that prohibition of deficiency judgments in some Canadian provinces increases the incidence of default. 7 Note that this expression differs from (5) only in that the lower bound of the integral is B rather than A, where B>A reflects the prohibition of deficiency judgments in this case.8 Thus, the integral term in (5) is larger than the corresponding term in (7), implying that mN<mD<m* , or, the borrower invests in a lower level of maintenance in the absence of deficiency judgments, all else equal. The loss in the value of the house due to reduced maintenance (given P) is given by Ph(mD)-Ph(mN).9 Note that this reduction in value reinforces the higher probability of default in the absence of deficiency judgments. C. Impact of Interest Rate Adjustments To this point, we have assumed that the interest rates were the same regardless of whether or not deficiency judgments were allowed. This allowed us to isolate the impact of differences in mortgage laws on borrower maintenance without worrying about the effects of interest rate adjustments aimed at equalizing lender profits. In this section, we briefly examine the nature of these second-order effects and show that they reinforce the first-order effects derived above. We first examine the impact of changes in the interest rate on borrower maintenance in the two cases. From the implicit function rule, we know that ∂mi / ∂r = (∂ 2 EU i / ∂mi ∂r ) /( −∂ 2 EU i / ∂mi2 ) , i=D, N, where the denominator of this expression is positive by the second-order condition. It follows that sign(∂mi / ∂r ) = sign(∂ 2 EU i / ∂mi ∂r ) . Differentiating (5) and (7) therefore shows that ∂mi < 0, i = D, N . (8) ∂r 8 The kink in the borrower’s wealth in Figure 2 thus corresponds to B. 9 Stein (1998: p. 1244) proposes a rule that would allow lenders to recover this loss in the event of default under non-recourse commercial loans. 8 Thus, maintenance is decreasing in r in both cases. Intuitively, an increase in r increases the range of borrower limited liability (i.e., it makes a deficit more likely), thereby reducing the marginal benefit of maintenance. Now consider the expected return to lenders under the two types of mortgages. When deficiency judgments are allowed, the lender’s expected revenue is A RD = ò [ y 2 + Ph( m)]dF ( P ) + [1 − F ( A)]L(1 + r ) . (9) 0 Similarly, when deficiency judgments are prohibited, the lender’s expected return is B R N = ò Ph(m)dF ( P) + [1 − F ( B)]L(1 + r ) . (10) 0 Differentiating these expressions with respect to r and m yields the following partial effects: ∂Ri ∂Ri > 0, >0 i=D, N. (11) ∂r ∂m Thus, lenders’ returns are increasing in the interest rate and borrower maintenance in both cases. In an equilibrium in which lenders are willing to offer both types of mortgages, the returns must be equal.10 Using (9) and (10), we find that B RD − RN = {ò [ L(1 + r ) − Ph( m)]dF ( P ) + F ( A) y 2 } > 0 . (12) A Thus, the return is higher when deficiency judgments are allowed, holding r and m fixed. This reflects the advantage to lenders’ of having recourse to borrowers’ non-housing wealth in the event of default. This difference in lender returns, however, cannot persist in equilibrium, given a national market for mortgages. In order to achieve equality, suppose that we (arbitrarily) fix rD 10 This assumes that the costs of originating a mortgage are roughly the same regardless of the legal regime. 9 and, given (11) and (12), increase rN. From (8) and (11), however, there is a second-order effect through borrower maintenance, yielding a total effect on RN of dR N ∂R N æ ∂R N öæ ∂m N ö = +ç ÷ç ÷, (13) dr ∂r è ∂m øè ∂r ø which is ambiguous in sign. If we assume, however, that the maintenance effect is small, then (13) is positive,11 and equalization of the returns to the two types of mortgages implies that rN>rD. This conclusion is consistent with Meador’s (1982) finding that mortgage rates are higher in states that prohibit deficiency judgments (all else equal).12 Note that the interest rate effect therefore reinforces the earlier finding of a higher level of maintenance when deficiency judgments are allowed. Specifically, we showed above that mD>mN for given r. Thus, if we increase rN while holding rD fixed, mN will fall further below mD (given (8)). Thus, even allowing for variation in interest rates, the prediction that borrowers invest in greater maintenance when deficiency judgments are allowed remains unambiguous. IV. Empirical Results A. Data We use the American Housing Survey (AHS) to test the predictions of the model. The AHS is a biannual national survey of approximately 50,000 households conducted by the U. S. Bureau of the Census. It consists of a panel data set that follows selected housing units over time. In each survey year (1985 through 1995) the Census Bureau collects information about the neighborhood, the home, the household, and occupancy costs. Of particular interest for this study, the AHS asks 11 Suppose, for example, that lenders are competitive and set r to maximize the expected utility of borrowers subject to a profit constraint and the fact that borrowers choose m optimally given r. Forming the appropriate Lagrangian and differentiating with respect to r yields the first-order condition: ∂EU/∂r+(∂EU/∂m)(∂m/∂r)+λ[∂R/∂r+(∂R/∂m)(∂m/∂r)]=0, where λ is the Lagrange multiplier. Note that the second term equals zero by the Envelope Theorem (given optimal borrower maintenance). Thus, since ∂EU/∂r<0, the expression in brackets (which coincides with (13)) must be positive. 10 households to report how much they spent on regular maintenance, major repairs (e.g., new roof), and improvements (e.g., kitchen or bath upgrades) during the last two years.13 We use the data from five national surveys (1985 through 1993) and limit the sample to owner occupied single-family houses. 14 We further limit the sample to households with a single active first mortgage loan originated at the time the homeowner bought the house. In order to control for the differences in mortgage laws that vary by state, we limit the sample to houses in identified PMSAs.15 We further limit the sample to homes purchased since 1975 so that we can use the Freddie Mac MSA house price indices and interest rates at the time of loan origination. Using the PMSA identifier and State Legislative Topics (1992) Volume 3: Foreclosure, Bankruptcy and Late Payments, we identify the mortgage law applicable to each observation. After screening for complete information on relevant variables, we are left with a sample of 6,847 observations.16 B. Variables The literature on homeowner maintenance provides little guidance on the selection of explanatory variables in a model of maintenance expenditures. Dildine and Massey (1974) and Vorst (1987) study maintenance in the context of an optimal control problem -- i.e., selecting the optimal level of maintenance and holding period for rental properties. These models and earlier qualitative literature (see, for example, Muth [1969] and Lowry [1960]) suggest that the rate of 12 Also see Stein (1998: p. 1218) who notes that commercial lenders typically require a higher interest rate for non- recourse loans. 13 Regular maintenance expenses are reported for the most recent twelve months. Major repairs and improvements are reported as totals for the last two years (i.e., since the last survey). 14 We exclude condominium units because the survey does not report information on association spending on maintenance or project characteristics. 15 The AHS does not identify the state on all records. A subset of each year’s sample reports the PMSA within which the property is located. 16 In addition to deleting observations with missing date, we excluded very low value homes (<$5,000) and imposed "reasonability" screens on data describing the estimated house value and mortgage debt. For example, we excluded observations with estimated current loan-to-value ratios greater than 5 as likely containing data errors. 11 change in quality of the structure and neighborhood should influence maintenance expenses. It is also clear, however, that maintenance and value are jointly endogenous: maintenance expenditures affect value and value influences maintenance. The theoretical discussion above suggests the following three-equation simultaneous system relating maintenance, value, and mortgage rate: Maintenance=f(Value, Rate, Hm, Om, M) (14) Value=f(Maintenance, Hv, Rv) (15) Rate=f(Maintenance, L, Rr, M). (16) In each equation, the capital letters with superscripts indicate a vector of house characteristics (H), owner characteristics (O), regional and neighborhood characteristics (R), mortgage loan characteristics (L), and mortgage law variables (M). Table 1 presents the complete vectors for each group of variables. The superscript on each vector indicates that only selected variables from each group are included in the different equations. Equation (14) allows both value and the mortgage rate to affect maintenance. The interest rate is included because, as noted, a higher interest rate increases the range of limited liability for the borrower. The value equation is a traditional hedonic model based on the premise that the value of a bundled good reflects the product of attributes and shadow prices of those attributes. The dependent variable is the homeowner’s estimate of the current house value. 17 Maintenance enters as an offset to depreciation, and also because our measure of maintenance expenditures includes spending on additions. Under this view, 17 Kiel and Zabel (1999) report that, although homeowners tend to overvalue their homes by about 5%, the difference between homeowner estimates and sales prices are not correlated with characteristics of the house, its occupants or neighborhood. The home values reported in the AHS are top coded at the 97th percentile of the full sample in a given survey year. A total of 377 observations were top coded. 12 neither owner characteristics nor the mortgage rate on the loan used to purchase the house affects the value.18 The rate equation is based primarily on the term structure of interest rates at the time the loan was originated.19 This equation allows for anticipated maintenance to influence the mortgage rate. The latter effect is justified by the assumption that lenders implicitly make an estimate of how well a particular buyer will maintain property serving as collateral for a loan in assessing the overall riskiness of the loan. We include in Hm, the vector of house characteristics entering the maintenance equation, the borrower’s estimate of current house value, and the three categorical indicators of structure age. We expect higher maintenance expenditures to be associated with higher value and older homes. The vector Om includes measures of the age, income, wealth, education, experience with homeownership, and time in the house. If maintenance is a normal good, we expect to find a positive relationship between income and maintenance. We measure income using the total salary income reported for the two highest earners in the household and an indicator variable signaling that the head of household receives social security income.20 The AHS data provides very limited information about household wealth. We use two proxies: the number of automobiles and trucks owned by members of the household and whether or not the household maintains a savings account. As with income, we expect higher wealth to be associated with higher expenditures on maintenance. Homeowner age, education, and experience with homeownership may be correlated with unobserved income and wealth, but may also have a direct effect. We expect that older homeowners may be less able to maintain their homes, and first-time homeowners may be less 18 We excluded “seller” originated loans from our sample. All loans are made by third party financial institutions and therefore there is less risk that below-market rates are capitalized into the selling price. 19 The month of origination is not recorded in the AHS data for loans originated in the period from 1975 through 1978. The origination month was set to June for these observations. 13 informed about good maintenance practices. Tenure in the home could affect maintenance if recent buyers are “house poor” and have limited discretionary resources for maintenance.21 Tenure in the current home is measured with two categorical variables indicating time in house of 1-2 years and 2-5 years. The excluded category is longer than five years. The only loan characteristic included in Lm is the interest rate on the loan. We do not include any regional or neighborhood variables in the maintenance equation, although these variables can influence maintenance indirectly through their effect on value. The vector M includes an indicator variable identifying those properties that are in states where deficiency judgments are granted when the lender follows the normal foreclosure procedure.22 Our theory indicates that when lenders have recourse to other assets of the borrower in the event of default, the borrower bears more of the true cost of overutilization and will tend to spend more on maintaining the property. We also control for borrower equity with an indicator that takes on a value of one when the borrower’s current loan-to-value ratio exceeds 90%.23 We estimate the current market value of the home using the original purchase price and the change in the Freddie Mac house price index for the PMSA. This estimate of value (and the resulting indicator value) avoids the potential endogeneity between homeowner-estimated value and past maintenance expenditures. We use the original loan amount24 as the numerator to estimate the 20 We do not have complete data on other sources of income or the amount of social security income. 21 The transfer of the property and the associated new mortgage provides an opportunity to finance repairs if buyers demand that sellers bring the home to a standard quality level as part of the sale negotiation. 22 Some states, including California, do not allow deficiency judgments unless the lender uses a judicial procedure instead of the standard power of sale process. The judicial procedure is rarely used, and California is categorized here as a non-recourse state based on the normal procedures used by lenders. 23 After allowing for normal sales and closing costs, owners with loan-to-value ratio greater than 90% have essentially no recoverable equity in the house. 24 The AHS data does not include current loan balance nor does it provide sufficient information to calculate a pro- forma balance without a significant loss of sample size. 14 current loan-to-value ratio. When the loan-to-value ratio is low, the borrower bears almost all the costs of overutilization, regardless of the local mortgage law. Although the vectors of house characteristics and regional variables included in the value equation are drawn from the list in Table 1, the specific elements are different from those in the maintenance equation. Following the customary structure of hedonic models, Hv includes the size of the home (measured in square feet), the number of rooms, and the number of bathrooms in addition to the categorical age variables. Demographic variables are generally not included on the right hand side of traditional hedonic equations and so are excluded here. We control for local price differences using a set of PMSA indicators. We suppress the coefficients for these indicator variables in our results. Rm also includes an indicator of dissatisfaction with the neighborhood on the premise that local amenities are capitalized into house prices, and an indicator that the property is in a location designated as a central city. A sequence of indicator variables for the different years is used to capture shifts in shadow prices over time. The interest rate equation is based primarily on the assumption that mortgage rates track market conditions. Consequently, Lr includes the ten-year government bond rate in the origination month and the yield curve slope (measured as the ten-year government bond rate less the three- month bill rate). In addition, Lr includes an indicator that the original loan to value ratio exceeded the 80% threshold at the time of origination.25 We include three regional indicators in Rr because there is evidence of regional variations in rates during the period these loans were originated.26 25 The original loan-to-value ratio used in this equation is based on the purchase price of the home and not the current estimate of value that was used in the current loan-to-value ratio in the maintenance equation. We use 80% as the threshold in the rate equation because the secondary market for mortgages traditionally differentiates high risk and low risk mortgages using the 80% threshold. 26 Longbrake and Peterson (1979) reported that mortgage rates in the 1970s averaged approximately 100 basis points lower in the Northeast region than in the West and South regions. More recent studies (e.g. Jameson, Shilling and Sirmans [1990] and Jud and Eppley [1991]) suggest that while the growth of the secondary market has reduced regional differences, local conditions remain significant in determining mortgage rates. 15 C. Results Table 2 reports our results. Three stage ordinary least squares was used for the estimation.27 The natural log of maintenance expenditures28 and house values were used throughout. The house characteristics enter the maintenance equation with the expected signs and significance. Maintenance expenses are higher with higher valued and older homes. Higher income, higher wealth, and more education are associated with higher maintenance expenditures. First-time homeowners and those who have just moved into their homes spend less on maintenance. The coefficient on the mortgage rate is negative, as predicted, but not significant. The imprecision in measuring this effect is not surprising given the small effect the mortgage rate, r, has on the range of limited liability and is consistent with the expectation that the rate effect would be smaller than the deficiency-judgment effect. The mortgage law effects are reported near the bottom of Table 2. As expected, homeowners in states that permit deficiency judgments spend approximately 19% more ($ 470) on maintenance of their homes than do homeowners in non-recourse states. The sign on the recourse indicator variable is positive and significant at the 5% level. The coefficient on the indicator of no recoverable equity is negative and very significant. This is consistent with the earlier results of Harding, Miceli and Sirmans (1999) that when homeowners do not expect to enjoy the future benefits of maintenance, they spend less. The other equations are of less direct interest but are presented here for completeness. The results in the value equation are generally consistent with previously published hedonic results. House value and size are positively associated. The PMSA indicators are jointly very significant, and prices are generally lower in central cities and unsatisfactory neighborhoods. As 27 The first stage regression results are not reported here but are available from the authors. 16 predicted, the coefficient on maintenance is positive and very significant. The mortgage rate equation shows that the government bond rate is the major factor determining mortgage rates. The sign on loan-to-value is positive, consistent with theory. The regional variables indicate that during this time period mortgage rates were somewhat lower in the East than in the other regions. The negative coefficient on maintenance is consistent with the expectation that lenders charge lower rates when the borrower is expected to maintain the house at a higher rate. Overall, the empirical results confirm that maintenance expenditures, value, and mortgage rates form an interrelated system. Further, the estimated coefficients on the mortgage law effects are consistent with the theoretical prediction. V. Conclusion This paper has presented a theoretical and empirical analysis of the impact of variation in mortgage laws on household maintenance decisions. The specific law we examined concerned the right of lenders to seek deficiency judgments against a borrower’s non-housing wealth in the event that the house value falls below the mortgage balance. The theory predicted that homeowners would invest in greater maintenance when lenders have this right. The reason is that homeowners have more to lose in the event of a deficit, so they will invest in greater maintenance in order to reduce the likelihood that the market value of the house will fall below the mortgage balance. We tested this prediction using data from the American Housing Survey and information on cross-state variation in mortgage laws. Because maintenance, house value, and mortgage rates are inter-related, we estimated a system comprising three simultaneous equations using three stage OLS. The results showed that homeowners in states where deficiency judgments are 28 In calculating the natural log of maintenance expenses, we added $1 to all reported amounts to assure that the dependent variable is well defined. 17 allowed invested at a higher rate, and the effect was significant, thereby confirming the predictions of the theory. 18 References Burke, D. (1989) Law of Federal Mortgage Documents, Boston: Little-Brown. Dildine, L. and F. Massey (1974) “Dynamic Model of Private Incentives to Housing Maintenance,” Southern Economic Journal 40: 631-639. Gibson, F., J. Karp, and E. Klayman (1992) Real Estate Law, 3rd Edition, Chicago: Dearborn. Harding, J., T. Miceli, and C.F. Sirmans (1999) “Do Owners Take Better Care of Their Housing than Renters?” manuscript, Department of Finance, University of Connecticut. Henderson, J.V. and Y. Ioannides (1983) “A Model of Housing Tenure Choice,” American Economic Review 73: 98-113. Jameson, M., J. Shilling, and C.F. Sirmans (1990) “Regional Variation of Mortgage Yields and Simultaneity Bias,” The Journal of Financial Research 13(3): 211-219. Jones, L (1993) “Deficiency Judgments and the Exercise of the Default Option in Home Mortgage Loans,” Journal of Law and Economics 36: 115-138. Jud, D. and D. Eppley (1991) “Regional Differences in Mortgage Rates: An Updated Examination,” Journal of Housing Economics 1: 127-139. Kiel, K. and J. Zabel (1999) “The Accuracy of Owner-Provided House Values: The 1978-1991 American Housing Survey,” Real Estate Economics 27: 263-298. Longbrake, W. and M. Peterson (1979) “Regional and Inter-regional Variations in Mortgage Loan Rates,” Journal of Economics and Business 31:75-83. Lowry, I. (1960) “Filtering and Housing Standards: A Conceptual Analysis,” Land Economics 36: 362-370. Meador, M. (1982) “The Effects of Mortgage Laws on Home Mortgage Rates,” Journal of Economics and Business 34: 143-148. Muth, R. (1974) Cities and Housing, Chicago: Univ. of Chicago Press. State Legislative Topics (1992) Volume 3: Foreclosure, Bankruptcy, Late Payments, L. McKenna, ed., Washington, DC: Mortgage Bankers Association. Stein, G. (1998) “The Scope of the Borrower’s Liability in a Nonrecourse Real Estate Loan,” Washington and Lee Law Review 55: 1207-1284. Vorst, A. (1987) “Optimal Housing Maintenance Under Uncertainty,” Journal of Urban Economics 21: 209-227. 19 $ Ph(m)+y2 Ph(m) Ph(m)+y2-L(1+r) L(1+r) Borrower y2 Lender Ph(m) Figure 1: Borrower and lender wealth when deficiency judgments are allowed. 20 $ Ph (m) Ph(m)+y2-L(1+r) L(1+r) Borrower y2 Lender Ph(m) Figure 2: Borrower and lender wealth when deficiency judgments are prohibited. 21 Table 1 Summary Statistics 6847 Owner-Occupied Single Family Houses from 1985-1993 AHS Category Variable Mean Std. Dev Min Max Maintenance Expenditures $2,467.81 $3,978.00 $0.00 $54,985.00 House Characteristics Homeowner Estimate of Value** $129,940 $78,336 $ 5,000 $ 350,000 % Value Top Coded 5.51% Age of Structure (yrs) 26.7 20.2 0 83 % 1-5 years old 10.44% -- -- -- % 5-10 years old 12.20% -- -- -- % 10-15 years old 13.52% -- -- -- Size (sq ft) 2,106 878 200 5000 Number of Rooms 6.8 1.6 2 16 Number of Bathrooms 2.2 0.9 1 12 Region & Neighborhood Characteristics % in Western Region 21.70% -- -- -- % in Southern Region 27.43% -- -- -- % in Midwest Region 28.30% -- -- -- % In Central City 34.56% -- -- -- % In Unsatisfactory Neighborhood 2.31% -- -- -- Homeowner Characteristics Age : Head of Household (yrs) 42.0 11.2 16 88 Salary Income $47,120 $30,425 $0 $200,000 % Receiving Soc Security 10.85% -- -- -- Mortgage Payment $ 744.94 $ 427.32 $ 52.00 $ 2,000.00 Education* % High School or Less 6.28% -- -- -- % College or More 46.69% -- -- -- Time in Home (yrs) 6.3 4.5 1 19 % In Home Less Than 2 Years 24.86% -- -- -- % In Home 2-5 Years 26.84% -- -- -- % First Time Owner 39.64% -- -- -- Wealth of Homeowner Number of Vehicles 2.1 0.9 0 8 Savings (1=Yes) 95.40% -- -- -- Loan & Market Characteristics 10 Yr. Treasury Rate at Orig. 9.18% 1.94% 5.33% 15.32% Yield Curve Slope 1.85% 1.30% -2.65% 4.42% Interest Rate on Mortgage 9.78% 1.69% 4.00% 20.00% % Loan-to-Value Ratio> .80 45.52% -- -- -- Incentives % with Loan-to-Value Ratio>.9 7.45% -- -- -- % Deficiency Judgment Allowed 78.94% -- -- -- * The excluded category is some college education without a degree. ** House value is top coded at the 97th percentile of the full year's sample. 377 observations were top coded. This table presents the summary statistics of a sample drawn from the 1985-1993 AHS surveys. The sample was restricted to owner occupied single family (attached and detached) houses where the owner also currently has an outstanding mortgage loan. The observations were distributed over time as follows: 1985- 1,593; 1987-1,184; 1989-1,447; 1992-1,319; 1993- 1,304. 22 Table 2 Three Stage OLS Estimation of Contractual Effects on Maintenance Expenditures Maintenance Value Interest Rate Coefficient t-statistic Coefficient t-statistic Coefficient t-statistic House Characteristics b ln(Estimated Value) 0.1741 2.01 a Top Code Indicator 0.3006 12.86 Age of Structure a a 1-5 years -2.6978 -23.27 0.4030 14.00 a a 5-10 years -1.1249 -10.83 0.1489 7.47 a b 10-15 years -0.5006 -5.05 0.0390 2.22 a Size (sq ft) 0.0001 17.47 a Number of Rooms 0.0725 17.62 a Number of Baths 0.1019 14.50 a a Maintenance Expenditures 0.0939 10.41 -0.1358 -7.99 Region & Neighborhood Characteristics West 0.0267 0.28 c South -0.0947 -1.77 Midwest 0.0330 0.64 a Central City Location -0.0511 -4.14 a Unsatisfactory Neighborhood -0.2665 -8.38 Year Indicator (1985=Base) a 1987 0.1184 7.73 a 1989 0.2407 16.51 a 1991 0.2720 18.17 a 1993 0.2544 16.12 Homeowner Characteristics a Age : Head of Household -0.0121 -3.42 a Salary Income ($,000) 0.0133 10.98 Education b High School or Less -0.2941 -2.39 a College or More 0.4881 7.49 Time in Home a 1-2 years -0.3118 -3.75 a 2-5 years 0.2788 3.53 a First Time Owner -0.5182 -7.65 a Receiving Soc Security 0.3737 3.39 Wealth of Homeowner Number of Vehicles 0.0559 1.64 a Savings (1=Yes) 1.1248 7.77 Loan & Market Characteristics a 10 Yr. Treasury Rate at Orig. 0.4058 43.16 a Yield Curve Slope -0.0456 -3.29 Interest Rate on Mortgage -0.0173 -0.51 -- -- a Original LTV Ratio > 80% 0.1686 4.58 Mortgage Law Incentives a Current LTV Ratio > 90% -0.4507 -3.93 b Deficiency Judgments Allowed 0.1867 2.19 0.0524 0.60 a a a Constant 3.1218 2.82 9.4907 86.88 6.8421 41.47 Number of Observations 6847 6847 6847 2 χ 1168.84 8582.86 1916.55 P-Value 0.0000 0.0000 0.0000 a=significant at the 1% level; b=significant at the 5% level; c=significant at the 10% level. Table 2 presents the results of three stage OLS estimation of the system of equations describing maintenmance, value, and mortgage rate. The natural log of maintenance and value are used throughout. Original LTV is the ratio of original loan amount to purchase price. Current LTV is the ratio of original loan amout to the estimated current market value of the house. A joint test of the PMSA indicators rejects the null hypothesis of no effect at the 1% level.

DOCUMENT INFO

Shared By:

Categories:

Stats:

views: | 6 |

posted: | 3/13/2011 |

language: | English |

pages: | 23 |

OTHER DOCS BY ps94506

How are you planning on using Docstoc?
BUSINESS
PERSONAL

By registering with docstoc.com you agree to our
privacy policy and
terms of service, and to receive content and offer notifications.

Docstoc is the premier online destination to start and grow small businesses. It hosts the best quality and widest selection of professional documents (over 20 million) and resources including expert videos, articles and productivity tools to make every small business better.

Search or Browse for any specific document or resource you need for your business. Or explore our curated resources for Starting a Business, Growing a Business or for Professional Development.

Feel free to Contact Us with any questions you might have.