System and method for providing property value estimates

[00002] This application is based on a provisional application, Ser. No. 60/056,196 filed on Aug. 21, 1997, the contents of which are hereby incorporated by reference. This application is also a continuation-in-part of U.S. patent application, Ser. No. 08/730,289 entitled METHOD FOR COMBINING HOUSE PRICE FORECASTS, filed Oct. 11, 1996, which is hereby incorporated by reference.

[00003] The present invention relates generally to estimating property values, and more particularly, to providing preprocessed property value estimates.

[00004] Financial institutions and businesses involved with sales of property have long tried to estimate values of property accurately. Accurate estimation serves many important purposes. For example, financial institutions use property value estimates as one of the key factors in approving mortgage applications for real estate sales. Relying on the soundness of the estimate, financial institutions accept the risk of lending large sums of money and typically attach the property as security for the transaction. Accordingly, the accuracy of estimated value of the real estate entity is critical.

[00005] In addition to the accuracy of the estimate, timeliness is a significant factor. For example, a closing on a real estate sales contract may depend on the buyer successfully obtaining a loan within a limited time period. Also, the ability to evaluate the value of a large number of properties in a short time frame is a business necessity. Bids on pools of seasoned loans, for example, are due in a very tight timeframe and require collateral evaluation. Hence, the ability to estimate the value of the real estate entity quickly is very important to lenders and prospective buyers.

[00006] According to current industry practice, an estimate for a particular property is produced by a system in real-time, but this poses many problems. For one, the process is slow, especially for processing bulk transmissions. The system also needs to validate the address of the property to be estimated to ensure that the input address is indeed a valid address. For example, the system may check against an existing database to determine whether the input address contains a valid zip code or whether a combination of city and state fields of the input address matches the zip code of the input address. Moreover, real-time processing is subject to unpredictable conditions such as availability of properly running equipment, appraiser availability, and scheduling conflicts, whenever estimates are necessary. Further, a system computes an estimate for a property each time a request is submitted, even for properties that have been previously computed. Such redundancy wastes processing time and resources, results in repeated data entry, and increases the cost of the property valuations.

[00007] Furthermore, in existing systems, the platform providing the valuation estimates also needs access to “raw” bases. For example, the Hedonic model requires the capture of property level characteristic data and transaction value data to provide an estimate. This may involve either the use of large amounts of electronic data storage devices or connectivity with other systems, such as a data provider's system, which results in either additional cost and/or a reduction of reliability. In the case of the repeat sales model, after the growth rate table is created it is applied to the “seed value data set” to create, the forecasts.

[00008] Therefore, it is desirable to increase efficiency of property value forecasts by streamlining the forecasting process.

[00009] It is also desirable to provide a timely and reliable estimate of value.

[00010] Systems and method consistent with the present invention provide timely and accurate property value estimates by preprocessing the estimates and storing them for later access.

[00011] Specifically, a method consistent with the present invention of providing preprocessed property value estimates comprises several steps. Initially, the system accesses stored property data. The system then computes a property value estimates of the accessed property data. Thereafter, the system stores the computed property value estimates in a forecast data repository.

[00012] A system consistent with the present invention for providing preprocessed property value estimates includes accessing means, computing means, and storing means. The accessing means accesses stored property data with which the computing means computes a property value estimates of the retrieved property data. Thereafter, the storing means stores the computed property value estimates in a forecast data repository.

[00029] Reference will now be made in detail to the present preferred embodiment of the invention, an example of which is illustrated in the accompanying drawings. Where appropriate, the same reference numerals refer to the same or similar elements. The appended claims define the scope of the invention; the following description does not limit that scope.

[00030] FIG. 1 shows a property value estimation system 100 consistent with the present invention. System 100 includes a computer 110, a property database 120, a forecast data repository 130, and a user terminal 140. In one embodiment consistent with the present invention, computer 110 includes a mainframe for managing the property data and a UNIX machine for processing the property data to compute property value estimates. The mainframe may be an Amdahl processor primarily using tape storage in addition to having approximately 10 Gb of DASD (Direct Access Storage Device). The UNIX-specific instructions may be executed on a Sun SPARC® 1000e with the Solaris®2.5.1 operating system, three system boards, six CPUs, 768 Mb of memory, and 306 Gb of disk space. One skilled in the art may, however, use any computing system with adequate processing and memory capabilities.

[00031] A program 115 controls the overall process of computer 110 and may be written using such tools as COBOL II, SAS, and IBM utilities (Syncsort, JCL, FTP, etc.). To perform some standard functions, program 115 may use various commercial software such as Group 1 (Code 1 Plus v1.5 and Demographics v2.7), MathSoft (S-Plus v3.4), and the SAS Institute (mainframe and UNIX v6.12). The use of these programs will be explained in detail below.

[00032] Unlike previous systems that computed property value estimates on-line, system 100 offers an efficient and flexible way of providing property value estimates. System 100 provides a collection of preprocessed property value estimates, which can be accessed to retrieve estimates of a property quickly. Program 115 can compute the property value estimates using any one or a combination of methods.

[00033] FIG. 2 shows a schematic overview of one embodiment consistent with the present invention. System 100 first accesses an address batch 210 representing property data of a portion or the entire country stored in property database 120, and provides the property addresses and related data for various statistical models.

[00034] In one embodiment consistent with the present invention, program 115 computes the estimate of property values using two statistical models, i.e., Repeat Sales model 220 and Hedonic model 230, and combines the results to produce the best estimate. Other models such as HNC model 240 and Value Point model 250 can also be used to compute the property value estimates. An example of combining results of the statistical models is provided in the above-referenced related application Ser. No. 08/730,289 For explanatory purposes, the described embodiment will combine the property value estimates of Repeat Sales model 220 and Hedonic model 230. One skilled in the art may use any one or a combination of methods without departing from the spirit of this invention. Regardless of what model or a combination of models are used, the property value estimates are stored in forecast data repository 130 for later retrieval. One skilled in the art may also design a system to store the property value estimates in database 120, which stores the property data.

[00035] Property database 120 stores data representing address batch 210 including property data and sales transaction data for use in computing the property value 20 estimate. A property data table 300 of FIGS. 3a-3c illustrate a sample format of the property data. Property data table 300 is composed of a field name column 310 containing address and county information, type column 320 indicating whether the field name contains alphabetic characters or numbers, length column 330 indicating the size of the field, and details/comments column 340.

[00036] A sales transaction, data table 400 of FIG. 4 illustrates a sample format of the sales transaction data including a key column 410, a field name column 420, a type column 430, a length column 440, and a comments/value range column 450. Key column 410 indicates with a check mark whether program 115 can sort the records based on the corresponding field name. The property data can be created by entities maintaining system 100 or obtained from commercial data providers such as DataQuick according to a predetermined schedule. In one embodiment consistent with the present invention, system 100 receives property data on a monthly basis from the data provider, typically on tape drives. System 100 stores the property data in property database 120 and updates the property data according to a predetermined schedule, e.g., every three months, to computer 110. One skilled in the art may easily vary the frequency and the period of the predetermined schedule at any time. For example, the predetermined schedule may be based on a time period or an occurrence of a certain event. In another embodiment consistent with the present invention, system 100 uploads the property data to computer 110 on a monthly basis as new property data files are received. One skilled in the art may, however, easily modify the frequency of data updates and the storage location.

[00037] Forecast data repository 130 contains property value estimates for each property stored in property database 120. As will be explained in detail below, program 115 may compute the property value estimates using a single statistical model or a combination of models. Computer 110 can also periodically update property value estimates, for example, every three months or whenever the property data is updated.

[00038] User terminal 140 can be a dummy terminal or a personal computer connected to computer 110. Through user terminal 140, users can access preprocessed property value estimates for any given property. For example, FIG. 5 shows three sample screens of user terminal 140. An initial screen 510 presents a menu of options the user can choose from. If the user selects option “S” to initiate the Home Value Estimator, for example, user terminal 140 presents a greeting screen 520 that welcomes the user and gives an option to continue or exit to the main menu. If the user enters “Y” to continue, user terminal 140 presents a data entry screen 530 requesting information or address of property for which the user would like an estimate. Thereafter, computer 110 accesses forecast data repository 130 to obtain the property value estimate corresponding to the data entered by the user.

[00039] FIG. 6 shows a sample output screen containing data such as property ID information, property value estimate, price trend, recent sale data, and area map. One skilled in the art may, however, easily modify the input and output screens.

[00040] As mentioned above, the described embodiment will be explained in detail assuming system 100 combines the results of Repeat Sales model 220 and Hedonic model 230 for computing the property value estimates. Also for explanatory purposes, system 100 updates property database 120 and computes new estimates on three-month cycles.

[00041] FIGS. 7 and 8 illustrate processes, preferably performed by the mainframe portion of computer 110, relating to a data warehouse platform. Specifically, FIG. 7 is a flowchart illustrating the process for creating a new master property file for Hedonic model 220. Initially, program 115 retrieves and consolidates backup DP (data provider) files for months 1, 2, and 3 stored in property database 120 (step 705). Program 115 also copies into memory the previous quarter's master property file, which program 115 used for computing previous quarter's property value estimates (step 710). Program 115 then identifies counties contained in the new property file and replaces the existing property data for each county with the updates (step 715), thus creating a new DP master property file (step 720).

[00042] Program 115 then “scrubs” addresses for each record in the new master property file to obtain a standardized version of each property address (step 725). During the scrubbing process, program 115 may use commercial software, such as Group 1 's Code 1 Plus and Demographic Coding System. Program 115 inserts the scrubbed address information in a corresponding field of property data table 300. These steps are repeated until program 115 reaches the end of the master property file. Program 115 thus produces a scrubbed property flat file (step 730).

[00043] To accommodate the process of Hedonic model 230, program 115 cuts the single large scrubbed property flat file into smaller county-specific files (step 735). Specifically, program 115 reads the scrubbed property flat file and creates a county-specific version of the flat file for each unique county encountered. Program 115 repeats these steps until it reaches the end of the scrubbed property flat file. Program 115 then transfers the resultant county-specific versions of the scrubbed property flat file to the UNIX portion of computer 115 where model processing will occur. A commercial package, such as File Transfer Protocol “FTP” utility by IBM, may be used to transfer the file. If program 115 encounters any errors during the cutting, scrubbing, or FTP process, program 115 generates an error report (step 740).

[00044] FIG. 8 is a flowchart illustrating the process for creating a new master property file, preferably performed by the mainframe portion of computer 110, for Repeat Sales model 220. The processes similar to that of one shown in FIG. 7. Initially, program 115 retrieves and consolidates sales transaction files for months 1, 2, and 3 stored in property database 120 (step 805). Program 115 also copies into memory previous quarter's sales transactions file, which program 115 used for computing previous quarter's property value estimates (step 810). Program 115 then identifies and appends any new sales to an existing property record in sales transaction data table 400 (step 815). These steps are repeated for each property record until program 115 reaches the end of the sales transaction file, thus creating a new sales transactions file (step 820).

[00045] FIG. 9 is a flowchart illustrating transaction data creation performed for Repeat Sales model 210. In general, the process of FIG. 9 identifies redundancies between the data provider, and loan agency data, e.g., FM/FNM (Freddie Mac and Fannie Mae), to create the most accurate record for a given property. To do so, program 115 compares the new sales transaction file against agency data to identify and resolve data redundancies (step 905). Program 115 then scrubs the addresses contained within the file and merges the unique agency and data provider data to create a larger temporary sales transaction file (step 910). As mentioned above, scrubbing may be performed using commercial software such as Group 1's Code-1 Plus and Demographics Coding System. From the temporary sales transaction file, program 115 generates two separate output files: seed value file containing most current sales price (step 915) and repeat sales file containing the scrubbed property address with all repeat sales transactions (step 920). Tables 1010 and 1020 of FIG. 10 shows a sample layout of the fields contained in seed value file and repeat sales file, respectively. Program 115 transfers seed value file and repeat sales file to UNIX portion of computer 110 for processing (steps 925 and 930).

[00046] The next phase of program 115 relates to CAD (Collateral Assessment Development) processes. In one embodiment consistent with the present invention, program 115 operates on a UNIX platform with SAS software. As shown in FIG. 11, program 115 uses repeat sales file from FIG. 9 (step 1105) to run the Repeat Sales model 220 to determine aggregate property value activity (e.g., increase or decrease in value) by property types (e.g., condos vs: single-family) within zip codes and counties for a given period of time (step 1110). To do so, program 115 analyzes the repeat sales file and groups the records by property type for each unique zip code and county. Program 115 then calculates the aggregate appreciation/depreciation in property value for the grouped sample. Program 115 may also create plot of indices to enable visual, inspection for quality control using commercial software such as MathSoft's S-Plus. Program 115 then produces a GRT (Growth Rate Table) or GMT (Growth Multiplier Table) by zip code 1210 and a GRT by county 1220, a sample layout of which are illustrated in, FIG. 12 (step 1115). GRT 1210 and 1220, in conjunction with a property's seed value, provide a market-to-market value of the property within a specified period of time.

[00047] Program 115 takes GRT 1210 and 1220 and applies it to records contained in the seed value file (step 1120) to create a RS (Repeat Sales) forecast (step 1125). A detailed explanation of Repeat Sales model 210 is provided in the above-referenced related application Ser. No. 08/730,289. This step produces a seed value file with estimated current value based on repeat sales.

[00048] Program 115 also takes county-specific versions of the scrubbed property flat file (step 1130) and merges the DP (Data Provider) records with repeat sales forecast information (step 1135) to obtain all data provider/FM/FNM records with most recent sale price and property data, preferably by county (step 1140). Program 115 then estimates county level estimate using Hedonic model 220 (step 1145) for all properties in the above-referenced related application Ser. No. 08/730,289

[00049] Program 115 applies Hedonic model 220 to all properties in the county (step 1150) and produces a data provider/agency properties file at the county level with hedonic estimate of value (step 1155). Program 115 then combines the file with Hedonic model 230 estimates of value with seed value file containing Repeat Sales model 220 estimated current value and generates the combined LP (Loan Prospector) estimate (step 1160). A detailed explanation of combining the results of the statistical models is provided in the above-referenced related application Ser. No. 08/730,289 Program 115 applies previously stored forecast data records from forecast data repository 130 (step 1165) and updates forecast data repository 130 with new property value estimates (step 1170). FIG. 13 shows a table 1310 illustrating representative fields of the records stored in forecast data repository. Once the UNIX portion of computer 110 completes processing to obtain property value estimates, program 115 transmits table 1310 and GRT tables 1210 and 1220 to the mainframe portion of computer 110 for data delivery to third parties (steps 1175 and 1180).

[00050] As shown in FIG. 14, records stored in forecast data repository 130 (step 1410) may be delivered to commercial data providers (step 1415) or be used to generate a report for a specific property (step 1420). GRT tables 1210 and 1220 (step 1425) can also be used to generate a growth rate report (step 1430) or be delivered to third parties (step 1435).

[00051] Users at user terminal 140 can now immediately access property value estimates by entering the input information of the property to be estimated. Once program 115 receives a request from user terminal 140, program 115 can access the preprocessed property value estimate and display the results in an output screen such as one shown in FIG. 6.

[00052] Systems and methods consistent with the present invention improve the efficiency of estimates of value for real estate entities by preprocessing the estimates and storing them for later access. Such systems and methods periodically process estimates of property value and store the estimates in a central repository. The central repository is periodically updated with new and revised property estimates.

[00053] It will be apparent to those skilled in the art that various modifications and variations can be made in the systems and methods of the present invention without departing from the scope or spirit of the invention. Other embodiments of the invention will be apparent to those skilled in the art from consideration of the specification and practice of the invention disclosed herein. It is intended that the specification and examples be considered as exemplary only, with the true scope and spirit of the invention indicated by the following claims.