Quantcast

What’s a Warranty Worth? The Impact of Home Owner Warranties on Property Sales

November 27, 2004

abstract

Home owner warranties (HOWs) are insurance policies that supplement regular homeowner’s insurance. HOWs cover major systems (e.g., HVAC, plumbing, etc.) in homes and are usually purchased during the marketing of a property. This study examines the impact of HOWs on a property’s price and marketing time; the results suggest that the residential real estate market does not reflect premiums in price or marketing time. It seems likely that mechanisms in the real estate market, including the uniform residential appraisal process, serve the role traditionally ascribed to HOWs and negate the effect of HOWs on price. In addition, the results indicate an adjustment for the presence of an HOW in the residential appraisal process is not warranted.

When selling an existing property, a seller or the seller’s agent may take action to make the property more attractive to potential buyers. Allowing extra inspections (assuming the property possesses no “surprise” factors, such as defects in a major electrical system or HVAC failure) and providing documentation of upgrades, pest treatments, or repairs are low-cost ways for the seller to assuage potential buyers’ concerns regarding a property’s condition. In some markets, however, this type of reassurance may not be enough to distinguish one property from another; in highly competitive markets, sellers may wish to affix a home owner warranty (HOW) to the property.1 By providing an HOW to cover the property in question, the seller effectively warrants that the property is free of defects in major electrical, mechanical, and plumbing systems. When viewed in this light, the HOW can be seen as a positive element; one might assume that the presence of the HOW should, ceterisparibus, increase the price of a property and/or decrease its marketing time.

HOWs typically provide coverage for major systems, including interior electrical systems, ceiling and attic fans, interior plumbing systems, and interior plumbing fixtures. Additionally, many major appliances are covered: washers, dryers, ovens, ranges, vent hoods, dishwashers, trash compactors, microwave ovens, water heaters, garbage disposals, water softeners, central vacuums, and garage door openers. Many policies also include a motel benefit for the policyholder in circumstances that render the property unusable for a period of time. Typical coverage, which includes all of the above items, costs approximately $400 with a $50 service call fee.3 Replacement costs for typical systems covered are reported in Table 1.

It should be noted that the HOWs in the study sample were all purchased by the seller of the property. The fact that an HOW was installed was reported by the listing agent through the multiple listing service (MLS) as a feature of the property on the listing date; as such, properties with an HOW in place serve as a direct test of the hypotheses concerning sellers’ motivations to install an HOW and the effect these HOWs have on residential property prices in the sample area. Buyer-initiated HOWs are marketed in the study area and are available for installation on the closing date or upon occupancy, whichever comes first. These warranties cover a 12-month period thereafter.4

Table 1 Typical HOW Components Covered and Replacement Costs

What if the HOW were not meant to be a signal that a property is without defect, but a preventive measure meant to protect the seller or broker from future liability? In this instance, where the pool of prospective buyers takes the less than positive view of the HOW, the presence of the warranty should decrease property price and/or extend its marketing time.

The purpose of this article is to present competing views of the HOW, to couch these views in financial theory, to analyze the impact of the HOW on the pricing of properties and their marketing times using a detailed data set, and most importantly, to determine what adjustments, if any, should be made during the residential appraisal process when evaluating properties both with and without an HOW. This article discusses the literature and competing hypotheses relevant to the study, presents the data and the methodology used to analyze the data, and discusses the results and conclusions.

Literature and Discussion

This study examines three possible scenarios regarding the effect of an HOW in the residential real estate market: the HOW may be a positive factor (increasing selling price and/or decreasing marketing time), a negative factor (decreasing selling price and/or lengthening marketing time), or a nonfactor (neither increasing nor decreasing the selling price or marketing time). Which of these three possibilities holds depends on many factors, including buyer uncertainty over property condition and the perceptions of buyers in the study area with respect to an HOW. Each of these three possibilities is presented and empirical tests are performed to determine which of the scenarios plays out in the sample.

The first hypothesis (Hypothesis One) is grounded in the Nobel Prize-winning work by George A. Akerlof, who first modeled the case of “lemons” in the market for used automobiles.5 Akerlof’s study models the used-automobile market in an asymmetric information environment. In that market, the sellers have more information than the prospective buyers regarding the used-vehicles. In such a market, which may contain many defective members or “lemons,” good vehicles are undervalued, due to the uncertainty introduced by the lemons and the information asymmetry. The solution to the information asymmetry problem is to limit uncertainty through the offering of a warranty. In the case of automobiles, the seller may guarantee the vehicle to be free of defects for some period, during which time any problem with the vehicle may be resolved at the seller’s expense.6

The parallel between the market for used automobiles and the market for pre-owned residential properties should be straightforward. Pre-owned properties have potential for defect similar to that of used automobiles; however, the potential for loss by a buyer of a property with defective systems is much greater, not only in terms of expensive repairs, but also lost resale value. Introduction of an HOW by the seller, then, can be viewed as a signal of higher quality, where the seller is willing to provide the warranty to differentiate his or her property from other less desirable properties in the same market. The resulting lower uncertainty for buyers of warrantied properties should lead to higher prices and/or shorter time on market, all else being equal.

While Hypothesis One indicates that an HOW may be a sign of a higher-quality property, there exists another, alternative interpretation of the signaling effect of the HOW, referred to as Hypothesis Two. Hypothesis Two draws from anecdotal evidence gathered from discussions with members of the MLS Committee in the sample area; it suggests that the presence of an HOW is a signal that the property may have some potentially serious defect. In this case, the seller may have placed the warranty on the property to protect against future liability for defects, including repair expenses and legal fees.

It is important to note that the seller’s own uncertainty over the condition of the property and the seller’s desire to avoid future problems after closing could be the proximate cause for placing an HOW on the property. In this case, the buyer of the property may perceive the presence of an HOW as a negative signal and accordingly lower the offer price, resulting in a price discount for the HOW property. Where the seller chooses not to accept a lower price, Hypothesis Two suggests that the marketing times for these warrantied properties will be longer.

The third hypothesis (Hypothesis Three) suggests that properties with an HOW see neither a price nor a time-on-market effect. The absence of either effect may indicate that there are market mechanisms in place that mitigate the need for an HOW in the sample market. These mechanisms could include inspections conducted by the buyer, professional inspections on behalf of the buyer, disclosure statements, and uniform appraisals conducted on the behalf of lenders to verify property conditions per underwriting requirements. Buyer inspections and professional inspections would obviously give the buyer more insight into the property’s condition, and disclosure statements are the standardized mechanism for discovery. Uniform residential appraisals, designed to protect the interests of the lending institution, may also serve as a mechanism for detection. All or any of these mechanisms could reduce buyer uncertainty over a property’s condition to the point where the need for an HOW is eliminated. Here statistical testing should indicate no impact on the part of HOWs in either pricing or duration (time-on-market) models.

The study analysis presented here will serve as an empirical test of these three hypotheses. Using statistical techniques, a model will be developed for the relationship between presence of an HOW and selling price and the presence of an HOW and marketing time. As previously discussed, the statistical relationships uncovered in this examination will serve to identify which hypothesis regarding the HOW holds for the sample area.

Data and Methodology

Data

Initially, the data set included \all 2,716 conventional residential closings between January 1, 1998, and December 31, 1998, in Montgomery, Alabama. The Montgomery Area Association of Realtors’ MLS was the predominant data source, providing data on selling price, selling time, location, and most of the physical characteristics of listed properties. However, information on age and square footage of listed properties did not appear in the MLS listings when the data was collected and was obtained from the Montgomery County tax assessor’s office.7

In order to obtain a complete set of characteristics for each sale, observations that did not appealin both the Montgomery Area Association of Realtors’ MLS and the Montgomery County tax assessor’s databases were expunged. Obvious data-entry errors from the MLS database, such as a negative time on market, zero bedrooms or baths, etc., were removed as well. After these adjustments, the final sample included 1,549 residential closings for the period in question. A summary of the descriptive statistics is presented in Table 2.

Pricing Effect Methodology

Table 2 Summary Statistics for Entire Sample

In this, as well as the time-on-market model, the regressors LnAGE, LnSQFT, LnBED, and LnBATH act as continuous variables representing the property’s age, square footage, number of bedrooms, and number of bathrooms, respectively. Four high school zones serve as proxies for location; three indicator variables are included in the model: CARVER, LEE, and LANIER. The omitted school zone, Jefferson Davis High School (JD), appears in the constant term and serves as the basis against which the other school zone estimates are compared. Also included are indicator variables controlling for the three available types of parking: garage (GAR), carport (CPT), and driveway (DRIVE). GAR and CPT are explicitly included in the hedonic model. DRIVE is the omitted categorical variable here.

In order to control for varying levels of quality among the sampled properties, five quality variables are included in the models along with the more conventional regressors already discussed. The presence of a fireplace (FP), a garden bath (GB), a separate shower (SEPSHOW), an in-ground swimming pool (POOL), or a double oven (DOUBOVN) is characteristic of higher-quality properties in the sample area. In an earlier work using this same data set, Johnson et al find that the presence of exterior insulation and finish systems (EIFS] as a siding was a positive and significant regressor in explaining property price.8 Building on this previous research, the models in the current work includes EIFS as a predictor, where EIFS takes a value of one if the property has exterior insulation and finish systems siding, and zero otherwise.

Timing Effect Methodology

This work employs the Weibull distribution over traditional OLS modeling to investigate the effect of HOWs on property marketing because, in addition to the flexible nature of the Weibull mentioned above, it is not subject to many of the known deficiencies of OLS estimation of property marketing time. For example, OLS modeling assumes a normal distribution of the error term. However, the non- normality of the error term, when estimating property marketing time, makes the use of OLS estimation suspect because it can lead to biased coefficients and therefore provide erroneous results. An excellent presentation on OLS versus duration methodology can be found in Keifer.9

Empirical Results

Pricing Effect

The results of the pricing model (Model 1) are presented in Table 3. With the exception of the variable of interest, the independent predictors in the model are all statistically significant and possess the expected signs. The coefficient of the continuous regressor AGE is negative and significantly related to LnSP, indicating that older homes sell for less than newer homes do, on average. SQFT, BED, and BATH are all positive and significant when related to LnSP; larger homes, homes with more bedrooms, and homes with more bathrooms tend to sell at higher prices than their counterparts do.

Proxies for location are discussed next. The Jefferson Davis High School zone represents the basis of comparison for location in the pricing model. Anecdotal evidence garnered from local real estate professionals provides the rationale for this decision. These real estate professionals indicated that the general preference ordering of school zones in the study area is Jefferson Davis, Lee, Lanier, and Carver, with Jefferson Davis representing the most preferred school zone. The coefficients for LEE, LANIER, and CARVER should be negative and increasing in absolute magnitude, given that the real estate professionals were correct in their assessment. The results are exactly as expected, and the location proxies are all statistically significant when related to LnSP.

The “quality” variables included in the model represent several items thought to signal a high quality home in the study area. These variables include indicators for the presence of a fireplace, a garden bath, a separate shower, an in-ground swimming pool, and/or a double oven. The quality variables are all positive and significantly related to LnSP, consistent with the a priori evaluation. Homes with only driveway parking available provide the basis for comparison of parking type. These “driveway-only” parking properties should be the least preferable, while properties with garage parking should be the most preferable. Therefore, the expectation is that GAR and CPT should have positive, statistically significant coefficients, with GAR having a greater magnitude. The results are as expected. In addition, and as earlier studies with this data set have found, the presence of EIFS has a statistically significant and positive effect on LnSP.

Table 3 Results of Hedonic Pricing Model

Finally, the variable of interest, HOW, is examined. It is insignificant in predicting sale price, indicating that the presence of an HOW does not significantly affect the sale price of properties in the study area. This result clearly does not support either of the signaling hypotheses offered earlier. To the contrary, this result suggests that perhaps those market mechanisms discussed earlier are serving in the role traditionally ascribed to HOWs. Furthermore, there is no statistical evidence to suggest the need for an adjustment for the presence of HOWs in the residential appraisal process.10

Timing Effect

To begin examination of the timing effect, the basic Weibull model (results reported in Table 4) is used, with a focus on the duration dependence parameter, a. In the simple Weibull model, α = 1.1142, indicating that the sample exhibits positive duration dependence; that is, the probability of sale increases over time. Analysis of the predictors of marketing time shows that very few have a significant impact on TOM.

Properties with a greater number of bathrooms (LnBath) have longer marketing times, on average, than their counterparts with a smaller number of bathrooms. Homes in the Robert E. Lee High School zone (LEE) sell faster than those in the Jefferson Davis (JD) zone. Only one of the quality variables is significant to marketing time: SEPSHOW, the indicator for a separate shower and tub. Properties with separate showers have significantly shorter TOM than their complements. As in Johnson et al.,11 the presence of EIFS siding extends marketing time significantly. Newly constructed properties (NC) sell faster than their pre-owned counterparts do. The factor of interest, HOW, is statistically insignificant in explaining marketing time. Specifically, its p-value, unreported, is 0.4121.12 This result adds to the body of evidence supporting the third of the three competing hypotheses.

Table 4 Results of Weibull Duration Model

Conclusion

This investigation centers around home owner warranties and their impact on property price and marketing time. Three competing hypotheses are presented; Hypothesis One indicates that properties with an HOW in place should lower buyer uncertainty over property condition and therefore sell for higher prices and/or exhibit shorter marketing times. Hypothesis Two suggests that an HOW is a negative signal and thereby increases buyer uncertainty with property price and marketing time responding accordingly. Hypothesis Three suggests that there may be market mechanisms in place that supersede the need for an HOW and, therefore, would remove any pricing or time on market impact.

The empirical examination here implies that neither a property’s price or its time on the market is affected by the presence of an HOW. Therefore, the experiential outcome fails to buttress either Hypothesis One or Hypothesis Two. Since the data indicates that there is no difference between properties with and without the HOW, the HOW must be without value, at least in the sample area. This lack of value may be conceived in the level of transparency present in the local residential market as suggested by Hypothesis Three. This suggests that in a market with high-quality inspections, seller disclosure, and relatively cooperative sellers, one should expect a high level of informational transparency and, therefore, no real need for an HOW.

The study results support the findings of many prior works concerning residential market efficiency, as these studies have suggested that real estate markets are informationally efficient.13 The results here also suggest that the residential appraisal process is correct in its omission of the HOW from a property’s valuation and that, in fact, an adjustment for the presence of an HOW would be inaccurate. There is no pricing or time-on-market effect on value because the overall system of appraisal helps reveal imperfections in properties. The line of reasoning that motivates the existence of the HOW is, at present, purely conjecture. However, there must be some impetus for sellers to purchase HOWs, and given \that they do not influence price or marketing time, the existence of these instruments should be explored through further research.

1. The home owner warranty, hereafter referred to as HOW, is an insurance policy that serves as a supplement to homeowner’s insurance. It is not a builder’s warranty such as that included with newly constructed property.

2. The replacement value of these items in 1998 dollars, excluding the motel benefit, is well over $3,500.

3. Additional coverage is available to sellers of properties to cover HVAC systems for a fee of approximately $60.

4. These warranties typically cover the same items as the seller- initiated warranties, with additional coverage for swimming pool equipment ($150 additional premium), refrigerators ($25 additional premium), whirlpool ($50 additional premium), and well pumps ($60 additional premium). All additional coverage items carry a $50 service call fee for the reported premia.

5. George A. Akerlof, “The Market for Lemons: Qualitative Uncertainty and the Market Mechanism,” Quarterly journal of Economics (August 1970): 488-500.

6. For an alternative application of this type of theory applied to real estate, see Patricia M. Rudolph, “Will Bad Appraisals Drive Out Good?” The Appraisal Journal (July 1994): 363-366.

7. Variables such as interior condition, decorating, motivation, etc. were unavailable, as the residential closings used here were historical and anonymous. This information is not collected by agents when listing properties and is, therefore, not included in the MLS database employed in this study. Such information could be extremely valuable in analyses such as the one presented. This information, however, is normally not available to appraisers and analysts even though it may be critically important to many data sets and situations.

8. Ken H. Johnson et al. “Exterior Insulation and Finish Systems: The Effect on Residential Housing Prices and Marketing Time,” Journal of Real Estate Research 22, no. 3 (2001): 289-311.

9. Nicholas M. Kiefer, “Economic Duration Data and Hazard Functions,” Journal of Economic Literature 26, no. 2 (1988): 646- 679.

10. In addition to the basic OLS model, we tested for selection bias via the Heckman Two-Stage Process. No selection bias was found and, therefore, the results of this superfluous model are unreported. The Heckman Two-Stage Process is presented as the primary method for dealing with sample selection bias in: William H. Greene, Econometric Analysis, 3rd edition (Upper Saddle River, NJ: Prentice-Hall, Inc., 1998), 706-714; T. Dudley Wallace and J. Lew Silver, Econometrics: An Introduction (Reading, MA: Addison-Wesley, 1988), 277-281; and Jeffrey M. Wooldridge, Introductory Econometrics: A Modern Approach (Mason, OH: South-Western Publishing, 2000), 560-562.

11. Ibid.

12. As a cautionary step, we also tested for heterogeneity in the time-on-market sample via the Weibull with Θ-Heterogeneity Correction [see Wooldridge (Ibid) for a discussion of this model]. The results indicated “no heterogeneity,” making the model unnecessary; therefore, the results do not appear here.

13. See Dean H. Gatzlaff and Dogan Tirtiroglu, “Real Estate Market Efficiency: Issues and Evidence,” Journal of Real Estate Literature 3, no. 2 (July 1995): 157-189.

Additional Reading

Elder, Harold W., Leonard V. Zumpano, and Edward A. Baryla. “Buyer Brokers: Do They Make a Difference: Their Influence on Selling Price and Search Duration.” Real Estate Economics 28, no. 2 (2000): 337-362.

Johnson, Ken H., Randy I. Anderson, and Justin D. Benefield. “Salesperson Bonuses and Their Impact on Residential Property Price and Duration.” Journal of Real Estate Practice and Education 7, no. 1 (2004): 1-14.

Johnson, Ken H., Thomas M. Springer, and Christopher M. Brockman. “Price Effects of Non-Traditionally Broker-Marketed Properties.” Journal of Real Estate Finance and Economics, forthcoming.

Johnson, Ken H., Leonard V. Zumpano, and Randy I. Anderson. “Intra-firm Real Estate Brokerage Compensation Choices and Agent Performance.” The University of Alabama Economics, Finance and Legal Studies Working Paper Series, 2003.

Jud, G. Donald, Terry G. Seaks, and Daniel T. Winkler. “Time on the Market: The Impact of Residential Brokerage.” Journal of Real Estate Research 12, no. 3 (1996): 447-458.

Munneke, Henry J. and Abdullah Yavas. “Incentives and Performance in Real Estate Brokerage.” Journal of Real Estate Finance and Economics 22, no. 1 (2001): 5-21.

O’Hara, Maureen. Market Microstructure Theory. Cambridge, MA: Blackwell Publishers, Inc., 1995.

Turnbull, K.T., C. F. Sirmans, and John D. Benjamin. “Do Corporations Sell Houses for Less: A Test of Housing Market Efficiency. Applied Economics 22, no. 10 (1990): 1389-1398.

Zumpano, Leonard V., Ken H. Johnson, and Randy I. Anderson. “Internet Use and Real Estate Market Intermediation.” Journal of Housing Economics 12 (2003): 134-150.

by Sean P. Salter, PhD, Ken H. Johnson, PhD, and Randy I. Anderson, PhD

Sean P. Salter, PhD, is an assistant professor of finance at the University of Southern Mississippi in Hattiesburg, Mississippi, where he teaches courses in corporate finance, financial valuation, money and capital markets, and investments. Prior to joining the faculty at USM, Salter served on the graduate faculty at the University of Alabama, where he taught courses in corporate finance, financial markets, and real estate. He has authored articles published in various outlets including the Journal of Real Estate Research. Contact: salter@cba.usm.edu

Ken H. Johnson, PhD, is an assistant professor of finance at Auburn University-Montgomery in Montgomery, Alabama. His articles have appeared in the Journal of Real Estate Research, the Journal of Housing Economics, and the Journal of Real Estate Law. Prior to joining the faculty at AUM, Johnson was a practicing real estate professional in Montgomery, Alabama, for over a decade. Contact: ken.johnson@mail.aum.edu.

Randy I. Anderson, PhD, is a professor and Ryder Eminent Scholar of Real Estate Finance in the CBA Jerome Bain Real Estate Institute at Florida International University in Miami, Florida. He currently serves as the Executive Director of the American Real Estate Society. Prior to his appointment at FIU, Anderson was the William Newman Chair of Real Estate at the City University of New York- Baruch College in New York, New York. Anderson has published articles in Real Estate Economics, the Journal of Real Estate Finance and Economics, and the journal of Real Estate Research, among others. Contact: randyiner@aol.com.

Copyright Appraisal Institute Fall 2004




comments powered by Disqus