Document View

This study provides a detailed look at the mechanics of the auctions from the Department of Housing and Urban Development and demonstrates how auction prices may differ significantly from non-auction prices for the same properties. The results confirm the general notion that, on average, auction properties sell for less than market value, but also show that the discounts may vary significantly across markets.

For many people, the phrase "real estate auction" leads to fantasies about opportunities to purchase properties at deep discounts. Presumably, the only properties on the auction block are those that must be sold quickly due to some urgency on the part of the current owner. If the owner is truly desperate to sell, then auction participants may be able to acquire properties at lower prices than those quoted in private negotiations with less anxious sellers. The purpose of this study is to analyze prices at one such auction, and to compare these with projected market prices had the properties been sold in non-auction transactions.

Auctions of many different types of assets occur frequently around the globe and are sometimes fascinating events both to observe and to participate in. Despite their morbidity, auctions of the remaining possessions of famous (or infamous) people are sometimes considered newsworthy events, possibly because of the relatively high prices the items may bring. For real estate market analysts, auctions can be enlightening (especially those conducted in the English style of oral ascending bids) because they provide a first-hand opportunity to observe a price established in a competitive market.

Under certain conditions, prices at auctions should be equivalent to those in non-auction transactions. Vickrey1 suggests that even if the number of participants in an auction is small compared to the number of bidders interested in the object in a non-auction environment, competition between participants (including the seller) keeps the price at the same level it would be in non-auction situations provided the participants do not engage in collusion, side payments, communications, or signaling that could affect the auction price.

Several recent auction research studies have focused on the micro-structure of the price formation process in auctions for various assets. Studies by Ashenfelter,2 Ashenfelter and Genesove,3 De Boer, Conrad, and McNamara,4 Mayer,5 and Mayer,6 question whether prices of similar items sold in a sequential auction are affected by the order of sale-in other words, are prices systematically higher or lower at the beginning of the auction than they are near the end of the auction? Although Mayer's results suggest that order of sale has no impact on prices, the other studies suggest that order does matter and that prices tend to fall as auctions progress.

In this study, we evaluate prices observed at a recent Department of Housing and Urban Development (HUD) auction for single-family homes in south Florida to determine if the auction element resulted in prices higher or lower than those that in a non-auction environment. Our results indicate that prices observed at the auction are less than the prices that would have resulted in non-auction transactions, confirming the general notion that auction transactions display unique "conditions of sale" that would prohibit their use as comparable sales in non-auction situations.

The first section of this paper describes the mechanics of a recent auction held on behalf of HUD to divest itself of properties acquired through foreclosures on Federal Housing Administration (FHA) insured loans. The discussion includes details about the quality of legal title conveyed by HUD to purchasers, buyers' obligation to inspect the properties before the auction date, real estate agents' role in the auction process, earnest money requirements, and closing arrangements. The second section of the paper compares transaction prices from this auction with predicted market values of the auctioned properties. The final section presents the conclusions drawn from this analysis.

The Mechanics of a HUD Home Auction

As a result of foreclosures, HUD frequently assumes ownership and possession of properties pledged as collateral for FHA-insured mortgage loans. The FHA insurance is a program in which the borrower purchases insurance that protects the lender in case the borrower defaults on the obligation to repay the debt. If the borrower defaults on an FHA-insured loan and the lender acquires tide to the property through the foreclosure process, the lender then files a claim with the FHA for reimbursement of any losses and simultaneously conveys title to the property to HUD which in turn tries to sell the property (by listing it with a local HUD-affiliated broker) to recover as much of the claim as possible.

When several HUD unsold properties accumulate in a given geographic area, HUD may arrange for an independent auctioneer to conduct a public auction of the properties. The frequency of these auctions nation wide depends on the rate of default on FHA-insured loans that result in foreclosure and local market conditions that may hamper sale of the property by listing it with a HUD-affiliated real estate broker.

Details of a HUD Auction

On July 10 and 11, 1998, an auction for single-family homes owned by HUD in south Florida was conducted in Fort Lauderdale, Florida, at the Radisson Bahia Mar Resort Hotel. The auction began at noon on each of the two days and continued for approximately two hours until the last property on the block for that day was sold. The auction was organized by Larry Lathem Auctioneers, a professional auction company headquartered in Phoenix, Arizona. Each of the properties offered at the auction was sold to the highest bidder. On the first day of auction, 71 properties located in Broward and Palm Beach counties were sold and on the second day, 99 properties located in Miami-Dade county and various counties along the southwestern coast of Florida were sold.

Prior to the auction date, the auctioneer advertised the event publicly by direct mail, newspaper advertisements, and the Internet to attract bidders. A list of properties to be auctioned was provided before the auction date, and potential bidders were encouraged to visit and inspect the properties prior to the auction and to attend a buyer's awareness seminar. At the seminar, the auctioneer informed potential bidders of the following general facts about HUD auctions:

* Properties sold at HUD home auctions are delivered to the buyer with clear title.

* All auction sales by HUD are final and all homes are sold on an "as is" basis.

* HUD discloses whether the property is eligible for FHA-insured financing, whether the property was built before 1978 (an indicator of potential lead-based paint on the premises) and whether the property is located in a flood zone.

* To facilitate potential bidders' inspection of the homes on the auction list prior to the auction date, bidders are encouraged to contact local real estate agents who have registered with HUD to sell HUD-owned homes.

* To register as a bidder at a HUD home auction, potential purchasers must prove that they have either sufficient cash to complete the purchase or have a letter of pre-approved credit from a HUD-approved lender.

* All bidders are required to have cash or a cashier's check for $2,000 as earnest money for each property the bidder intends to buy. Unless HUD grants an extension, earnest money is forfeited if the purchaser fails to close on the property using a designated HUD closing agent within 30 days of the auction date.

Although these details pertain to the aforementioned HUD auction in Florida, the arrangements are similar to those at other HUD auctions nation wide. Of course, unique characteristics of this auction may limit the generality of our analysis. This paper presents an analysis of the transactions at this particular auction to provide insight into the auction price formation process.

Analysis of Auction Prices 3

For each property sold at the auction, the price was recorded in a database, along with the location of the property (address, city, county), number of bedrooms, number of bathrooms, whether the property was constructed prior to 1978, flood zone status, and whether the property was eligible for FHAinsured financing (Table 1).

Of the 131 homes in the databases, the average price of homes sold at the auction was $59,653, substantially less than the average selling prices of homes in this area. In comparison, the average price of homes sold in non-auction transactions during the month preceding the auction in the three counties involved (Broward, Miami-Dade, and Palm Beach) were $145,534, $184,263, and $196,404, respectively. The fact that the average price at the HUD auction is less than the market averages is not surprising because it may simply indicate that the homes sold at auction were less desirable (and valuable) those typically sold in these areas.

An interesting question is whether the selling prices from the auction would be the same for the same homes in a non-auction transaction. Since these are real properties and are not an experimental model, we can never know for sure how much the properties would have sold for in a non-auction situation. We can only use estimated prices for analysis. The method described below for calculating these estimates relies on assessed values as determined by the local property assessor.

Using Assessed Value to Estimate Market Value

Determining whether auction prices are systematically higher or lower than prices in non-auction situations requires an accurate estimate of the price each property would bring in a different market. Short of conducting a complete appraisal for each property, how can analysts estimate this amount?

One possible approximate of market value readily available to bidders at real estate auctions is the assessed value of the property as determined by the local property assessor. Indeed, conversations with several participants at the auction revealed that the assessed property value was the highest amount they would be willing to pay for any particular property, assuming the property was in reasonable condition. Using assessed value as a substitute for market value, however, raises several problems. First, assessments are performed only periodically (as of January 1 st annually in Florida) and may not accurately represent the value of the property on the date of the auction due to changes in market conditions or changes in the condition of the property. Second, there may be a tendency among assessment officers to make conservative estimates to reduce assessment appeals. Third, as Goolsby7 reported for owner-occupied homes in various counties in the state of Washington, the use of standardized adjustment premiums and discounts for certain property features by assessors may result in consistent biases in the assessment process. Fourth, as demonstrated by Sirmans et al,8 assessed values may not always be equitable across different value ranges of properties. For example, higher priced homes may be assessed at a higher (or lower) proportion of their market value.

Enlarge 200% Enlarge 400%

Enlarge 200% Enlarge 400%

Enlarge 200% Enlarge 400%

To address these concerns, we obtained predicted market value from assessed value for each property, i, by dividing that property's assessed value by the mean assessment ratio for recently sold properties from each geographic area of our sample, If the concerns discussed above are economically significant, we expect their effects to be standardized for the mean property. Predicted market value for property i is thus calculated as:

Enlarge 200% Enlarge 400%

Enlarge 200% Enlarge 400%

The control sample was screened to eliminate transactions in which the sale price negotiations for a single property were atypical (i.e., sales between related parties, sales involving corporate buyers or sellers, sales involving lending institutions, sales involving more than one tax parcel).

Table 2 provides summary statistics for the data used to calculate the mean county assessment ratios. The mean assessment ratios for sales during the month preceding the auction for Broward County, Miami-Dade County, and Palm Beach County are 82.4%, 95.1%, and 79.4%., respectively.

The application of the ratios to obtain predicted market values as defined above is shown in Table 3. The discounts from market value range from 13.34% in the Miami-Dade county sub-sample to 21.54% in Palm Beach county. For the full sample of properties (131 total), the results suggest that auction prices show an average discount of rate 17.5%.

To determine if the difference between auction prices and market values is statistically significant, we employ a paired t-test for each sub-sample and the full sample. The test statistic is given by:

Enlarge 200% Enlarge 400%

The statistic is distributed as Student t-test with n - 1 degrees of freedom, and the null hypothesis of "no significance" is rejected when the statistic is greater than the corresponding critical value from the t-table. This paired t-test assumes that the population of the differences is normally distributed. Higher values for the test statistic imply greater statistical significance, with any value greater than 2.0 suggesting strong statistical significance. The results shown in the bottom row of the table strongly indicate that the discount is statistically significant for the full sample and for all sub-samples considered.

Conclusions

The perception that auctions present buyers with extraordinary opportunities to acquire assets at deep discounts may be fueled by the notion that assets are placed for auction when owners face an urgent need to dispose of the assets regardless of price. The results presented in this study demonstrate that auction prices are consistently below predicted market values, although the amount of the discount varies across the geographic sub-samples considered. Although the purpose of HUD auctions is to dispose of property in a timely manner, it cannot be assumed that all HUD properties are in distressed markets. The auction detailed herein, held on July 10 and 11, 1998, occurred in the framework of robust national and local economies.

This study reviews auctions of HUD properties acquired as a result of borrower defaults on FHAinsured home mortgages. Using data from transactions at a HUD auction held in south Florida, we considered whether prices from the auction differed significantly from predicted market values. Although our method of predicting the non-auction prices is certainly less than perfect, it contains data widely available to the typical auction participant, and therefore may well reflect participants' rationale in determining value estimates for properties sold at real estate auctions. The results of our analysis support the general contention that auctions present unique 19 conditions of sale," and that auction prices are likely to result in price discounts that may vary significantly from one real estate market to another. In this study, we found discounts ranging from 13.3% to 21.5% across our geographic sub-samples, with an average discount of 17.5%.

[Author Affiliation]

Marcus T. Allen, PhD, holds a doctorate from the University of Georgia's Terry College of Business. In addition to teaching in the real estate program at Florida Atlantic University since 1992, he has published several textbooks and has conducted numerous research studies that have been published in various real estate and finance journals. He maintains an active consulting practice and is a registered appraiser and licensed salesperson.