AU2003203821A1 - Valuation exchange - Google Patents

Description

AUSTRALIA

Patents Act 1990 RESIDEX PTY LIMITED COMPLETE SPECIFICATION STANDARD PATENT Invention Title: Valuation exchange The following statement is a full description of this invention including the best method of performing it known to us:- Title Valuation Exchange Technical Field This invention concerns the valuation of real estate. In particular it concerns a valuation exchange, and a method of operating such an exchange, to value individual real estate properties and to provide a valuation advice on request. In addition to the exchange performing an electronic valuation process it is able to attach a warranty to the valuation.

Summary of the Invention In a first aspect, the invention is a computerized valuation exchange to receive valuation requests in relation to properties within a region, and to provide a valuation advice in response to each request. The exchange comprising: A data input port to receive data about property sales within the region, the data including the date and value of property sales that have occurred in pre-determined geographical areas of the region; and to receive valuation requests for properties within the region.

A processor to: Calculate statistics from received property sales data.

Derive capital growth indices for each area based on the change in sale price of properties that are sold more than once.

Derive residuals to describe the distributions of deviations from the capital growth indices.

Derive the distribution of errors between purchase prices and a range of possible purchase prices, using the residuals. And, Estimate a range of property values for each property within the region using the statistics, indices, residuals and errors.

A computer memory to store data representing the determined range of property values.

A data output port to deliver a valuation in response to receipt of a valuation request at the data entry port.

Such valuations may be issued for properties without the need to inspect the properties themselves.

Where a valuation request occurs at the time of a purchase, the valuation request may include an advised value for the respective property. In this case the advised value may be taken into account when estimating the value.

The estimate may also take account of one or more of the following: Historic sales data for that property, the appropriate capital growth index and the residual; previously determined property values in the immediate vicinity of the property; and the statistics.

These may then be combined to give the range of property values. A value from the range, such as the median, may be taken to give a single determined property value.

In the situation where a financier provides a loan to a borrower for the purchase of a property, and the property is provided as a security for the loan.

The borrower may subsequently default on the loan repayments, and the financier may enter into a 'mortgagee in possession' sale to recover funds to clear the loan. In a small number of cases the sale does not recover sufficient funds to completely clear the loan. The valuation provided using the invention may be used to provide a guarantee (or warranty) of a value for the property at the time of the loan, to reduce the,, risk to the financier that the value of the property is insufficient to clear the loan. The warranty may be provided for a fee, and in the event that a 'mortgagee in possession' sale does not recover sufficient funds to completely clear the loan, a claim can be made against the warranty.

The valuation, or warranty, request may be made over the Internet to a web site that provides the data input and output ports. Alternatively, a phone may be used by either voice or messaging. Any other business communications technique or medium may also be used. The request may include the sale price or the amount of the loan, or merely the identity, such as the address, of the property.

The stored data typically includes the dates and values of any sales that have occurred, grouped by geographical area, for as long as sales records have been kept.

The determination of property values may take place on a regularly recurring basis, after new sales data is received, and requires no physical inspection of the property.

The calculation of a valuation, or a warranty fee, may take place immediately in response to receipt of a warranty request.

In a further aspect, the invention is a method of operating a computerized valuation to receive valuation requests in relation to properties in a region, and to provide a valuation advice in response to each request. The method comprising the steps of: Receiving data about property sales within the region, including the date and value of property sales that have occurred in pre-determined geographical areas of the region.

Calculating statistics relating to the property sales data in each geographical area.

Deriving capital growth indices for each area based on the change in sale price of properties that are sold more than once.

Deriving residuals to describe the distributions of deviations from the capital growth indices.

Deriving the distribution of errors between purchase prices and a range of possible purchase prices using the residuals.

Estimating a range of property values for each property within the region using the statistics, indices, residuals and errors.

Storing data representing the determined range of property values.

Receiving a valuation request at a data entry port.

Calculating a value automatically in response to a request, using the range of property values.

Outputting, at a data output port, the valuation and the warranty.

Brief Description of the Drawings An example of the invention will now be described with reference to the accompanying drawings, in which: Fig. 1 is a schematic diagram of a computer system.

Fig. 2 is a diagram of a web site page. And, Fig. 3 is a flow chart of a valuation and warranty fee calculation.

Best Modes of the Invention A financier 1 provides a loan 2 to a lender 3 for the purchase of a property 4. The property 4 is provided as a security 5 to the financier 1. The financier 1 is able to seek a warranty against subsequent loss using the valuation exchange computer system The financier 1 uses a computer 6 equipped with browser software to navigate across the Internet 7 to a web site 15 supported by a web server 11 in the computer system 10. The financier navigates to page 20 (see Fig. 2) and provides the purchase price, or a price estimate, by entering these amounts in respective data entry boxes 21 and 22, and requests 23 a fee advice for the warranty by submitting the completed page.

The system 10 will have determined its own value for all properties within the geographical areas for which comprehensive data is held, including this property. These values are stored in memory 16. The system 10 responds to the warranty request by using its already determined value together with the purchase price, or estimate, to calculate and issue a fee advice 24 for the cost of providing a warranty that a 'mortgagee in possession' sale will recover a guaranteed price related to the purchase price, or price estimate. The financier can choose whether or not to accept the fee and purchase the warranty.

Should a loan go into default and the property securing it is sold, then a check is made to determine whether the guaranteed price was recovered. If not, the financier may be compensated under the warranty.

Property Value Determination The system 10 relies upon determining its own value for the properties that it will insure. It determines values for each property every month. To do this the system 'engages in a number of activities, see Fig. 3: First, at the beginning of each month it collects reliable property sales data 31 from publicly available government sources. Typically, 10-20% of the data represents sales for the previous month, 60-70% represents sales for the month prior to that, and the remaining sales data represents earlier months.

These data are added to the system databases 16 and are then available to be used to determine property value data.

Second, a variety of statistics relating to the sales data are derived 32.

These include the median, and the upper and lower quartile sale prices for sales in each postcode area.

Also, capital growth indices are derived 33 for system-chosen small geographic areas using a repeat sales methodology, to update stored property prices in each area. This is done for example, by constructing matrices representing earlier and later sale prices, growth indices and errors, and by minimising the sum of squares of errors, or by any other known mathematical technique, to derive a matrix of growth indices.

This technique has been found to work well for areas where there are large numbers of sales. However, for smaller areas the results can be volatile.

The technique can be adjusted if necessary to address this volatility. One adjustment uses a larger area 'base' index as a weighting factor 34 in the index for a small area index. The 'weighting' factor determines how heavily weighted towards the base index the resultant index will be. A weighting of zero implies that the base index has no influence on the result. As the weighting approaches infinity, the index approaches the base index precisely. The weighting used in any particular situation can be empirically derived.

Residuals 35, that is the distributions of deviations from the capital growth indices. The residuals tell us the extent to which the growth rate of a pair of sales on an individual property is likely to deviate from the average growth rate measured by the index.

The distribution of errors on purchase prices 36. That is, the amount by which a property's sale price deviates from its value. The sale price of a property is dependent on a number of factors, some fundamental to the property, such as its location, quality and condition. Others relate to the buyer and seller, such as negotiation skills, urgency for a sale and urgency for a purchase. The latter factors mean that the same property at the same point in time could sell for over a range of different prices.

Because only one sale ever takes place at a single point in time, the existence of a sale price distribution is a subtle point often missed. The consequence of the sale price distribution is that the sale price at a point in time is not the 'value' of the property at that time (as is often assumed). The sale price is a single observation of the distribution centred around the value of the property. The difference between the sale price and the value (the median of the distribution) is called the sale price error.

The difference between these two interpretations manifests itself when we consider repeat sales of properties. On average, we expect the ratio of the two sale prices to be in line with the appropriate house price index for the area over the same time period. However, not all properties will move in line with the average and the deviation from the index can be measured for a variety of different time periods. If we allow for sale price errors, then there will be an error distribution at the initial sale, and another error distribution at the second sale. Both must be included as possible sources of deviation from the capital growth index. The two sale price errors are independent of the length of time for which the property was held.

In contrast, the longer the time period for which the property has been held, the greater the potential for deviation from the average growth rate, for instance because there are more opportunities for significant alteration to the property. This is quantified by the residuals and the residuals are calculated for each pair of sales on an individual property. The residuals are then grouped according to how long the properties were held. From this data a statistical extrapolation is able to locate the likely deviation if the pair of sales had occurred on the same day. This deviation represents the sum of the two sale price errors Third, the system estimates the value range for each property in up to three different ways: The first estimate 37 is based on historic sales data. If the data includes an historical sale of that property, then the appropriate capital growth index is applied to update the historical sale data to a current day value. By applying the residual a range is derived for the current day value of the property from the historical sale.

Where there is more than one historical sale, then the range generated from each sale are weighted together to form an overall range based on capital growth. This weighting is based upon a statistical confidence that has been calculated to determine the likely correctness of each of the historical sales at the time of that particular sale.

Where there are no historical sales on the database for the property, then this price estimation step is omitted. This results in a decrease in the accuracy of the resulting combined estimate.

The second estimate 38 of the value of the property is based upon previously determined property values 39 in the immediate vicinity of the property.

Typically, this means properties in the same street, but where a street is so small that it contains fewer than, say, twenty properties of the same type (house or unit) as the one being examined then the focus is widened, street by street, until a statistically sufficient number of properties have been identified.

The previously determined values are those determined for the previous month.

As each property used in the immediate vicinity has a price range attached to it, the range for the property is found by summing the range of values in the immediate vicinity. The summing is weighted by a statistical confidence measure in the values.

The final estimate 40 of the property's value is based on the postcode statistics derived at above. Assuming there are sufficient sales say more than twenty, the range of recent sales prices is used as the third estimate of the possible property value. If there are insufficient recent sales in the postcode, then the process uses the statistics for a specified larger region such a group of postcodes.

Having generated up to three estimated ranges of property values for the given property, it is then necessary to combine them into one range and then to determine a property value 41. To do this the three ranges are multiplied together, with appropriately calculated statistical weightings. The median of this range is taken to be the determined property value.

Each month determined property value ranges and the underlying distributions are 'calibrated' 42 by comparing the determined property values to the sales that have occurred, and providing a correction. This ensures that the error distributions recorded in the database are reflective of the 'real world' accuracy.

Warranty Fee Calculation The advised purchase price or price estimate provided by the financier has two functions. First, it is the price for which the Warranty is calculated.

Second, it is an estimator of the property's value in its own right. However it is necessary to know whether the price was derived as a consequence of an arms length third party sale by contract, or, whether it was derived as a consequence of the borrower or some other party estimating its value. In either event it is possible to determine an appropriate distribution of error.

The system now has two estimators of the property's value the 'determined property value' and the advised price. Further, it knows the range of the determined property value and is able to calculate an error distribution for the advised value.

The system then combines the two estimators by weighting the two estimators together in inverse proportion to the variances of their range and distribution, to produce a combined value estimate (CVE) 43.

Other factors may be taken into account, such as: a loan to valuation ratio; a probability of default; a probable timing of any default; a risk of fraud; the state of the overall property valuation guarantee portfolio; A distribution of the timing of loan losses resulting from mortgagee in possession sales; A distribution of loan to property values used as securities for loan losses resulting from mortgagee in possession sales; A distribution of dollar value losses resulting from mortgagee in possession sales; A distribution of errors with respect to price estimates made by borrowers where the property is not subjected to an arms length third party sale; Reliable and robust capital growth indices for small areas of a city or town, as described above in and A distribution of errors with respect to the determined property values and actual market values.

The warranty fee is calculated 'on the fly' by the system, taking into consideration the amount being guaranteed, the CVE and an error distribution.

First, the expected loss 44 is calculated by the following formula: Expected Loss Prob. of payout x expected size of the payout The expected loss is not a single figure. It is a distribution of possible values. The process requires a point on the distribution to be chosen based on the size of the portfolio and the amount of risk the user is willing and allowed to take for the portfolio. This point will therefore be adjusted continually as the size of the portfolio changes and when and if clustering in any geographic region takes place.

The Warranty Fee 45 is then calculated so as to ensure expected losses are covered with a sufficient degree of confidence and a profit margin is attained.

Warranty Claims For a claim to be made against the Warranty, two events must occur: First, the loan must go into default and the lender must enter into a mortgagee in possession sale which, plus or minus agreed adjustments, must provide insufficient funds to clear the loan. The probability of the property going into default and the sale making a loss is a function of a number of things the credit risk of the borrower whom either purchased the property or made the price estimate, the LVR of the loan, etc. This is more or less a standard analysis of risk that a lender or mortgage insurance company daily takes and is relatively trivial to calculate given appropriate data.

Second, the valuation must be shown to be inaccurate at the date it was calculated, by indexing back from the mortgagee in possession sale price. This calculation will also be adjusted to take account of factors such as abnormal maintenance which could result in a decrease in property condition and resale value after the valuation took place.

The probability of the sale price being shown to be inaccurate is a function of the accuracy of the CVE and its relative value to the advised price. It is also a function of the length of time the property is held before its mortgagee in possession sale. And lastly, it depends on the sale price achieved at that mortgagee in possession sale relative to its true value at the time of the mortgagee in possession sale.

Payout Amount The size of the payout is the lesser of the loss made by the mortgagor and the apparent error between the advised price and its back-indexed value based on the mortgagee in possession sale price. It is assumed that this difference is paid out with the addition of an interest component for the time between the warranty being given and the mortgagee in possession sale As with the probability of a default, the magnitude of loss for the lender is a more or less standard calculation given appropriate historical data and is calculated in a similar manner to that which should be used by a mortgage insurer.

Example Property A has an advised price of $200,000. The loan amount is $180,000. The determined property value (RVE) is $210,000. Assume that the standard deviation of the advised price as an estimator of the value is 6%.

Assume that the standard deviation of the RVE distribution is Then, the Combined Value Estimate (CVE) is $210,000 A A 2 A 2 15% A x $200,000 A (15% A 2 A 2 15% A2) $201,350 The standard deviation of this estimate is given by x A 2 (15% A 2 6% A 5.57% If we assume that, on average, defaults take place after, say, 4 years and that the standard deviation around- the index increases by, say, 1% per year then the distribution of the back indexed value of the property will have a mean of $201,350 and a standard deviation of SQRT(5.57% A 2 6% A 2 (4 x 9.11% So, the expected perceived error should the loan default is distributed with a mean of -$1,350 and a standard deviation of $18,348 This translates to a truncated distribution with a mean of $6,758 and a standard deviation of $10,124 Assuming the loss for the lender will be larger than $6,758 and based on a default rate of, say, 2% for this particular loan-to-valuation ratio, this translates to an expected loss of $135 with a standard deviation of $202.

Assume that the required confidence level of avoiding loss for the portfolio is, say, 99.9% and assume that for the size of the portfolio, this translates into a confidence level for the individual loan of 52% then this will translate to a loss factor for this guarantee of approximately $145.

If we add a profit margin of, say, Then the guarantee/warranty fee amount charged is $195.

In practise, much of the calculations that need to be done can be done independently of the advised price. These calculations would be done prior, condensing most of the process to a series of lookup tables, produced each time the system recalculated the determined property value RVE.

So, for example, in practise, Property A would be put to the system.

For a sale price that deviated from the RVE by -4.76% with a RVE standard deviation of 15% and LVR of 90%, this gives a payout value of 0.07% of the property's sale price plus profit margin.

It will be appreciated by persons skilled in the art that numerous variations and/or modifications may be made to the invention as shown in the specific embodiments without departing from the spirit or scope of the invention as broadly described. The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive.

Claims (26)

1. A computerized valuation exchange to receive valuation requests in relation to properties within a region, and to provide a valuation advice in response to each request, the exchange comprising: a data input port to receive data about property sales within the region, the data including the date and value of property sales that have occurred in pre-determined geographical areas of the region; and to receive valuation requests for properties within the region; a processor to: calculate statistics from received property sales data; derive capital growth indices for each area based on the change in sale price of properties that are sold more than once; derive residuals to describe the distributions of deviations from the capital growth indices; derive the distribution of errors between purchase prices and a range of possible purchase prices, using the residuals; and, estimate a range of property values for each property within the region using the statistics, indices, residuals and errors; a computer memory to store data representing the determined range of property values; and, a data output port to deliver a valuation in response to receipt of a valuation request at the data entry port.

2. A computerized valuation exchange according to claim 1, wherein the data input port receives a valuation request at the time of a purchase, the valuation request includes an advised value for the respective property, and the processor takes account of the advised value when estimating the value.

3. A computerized valuation exchange according to claim 1 or 2, wherein the processor also takes account of one or more of the following factors: historic sales data for that property, the appropriate capital growth index and the residual; previously determined property values in the immediate vicinity of the property.

4. A computerized valuation exchange according to claim 3, wherein the processor uses a combination of the factors to estimate the range of property values.

5. A computerized valuation exchange according to any preceding claim, wherein a value from the estimated range is taken to give a single determined property value.

6. A computerized valuation exchange according to claim 5, where the value is median.

7. A computerized valuation exchange according to any preceding claim, wherein the valuation is used to provide a guarantee (or warranty) of a value for the property at the time of the loan.

8. A computerized valuation exchange according claim 7, wherein the valuation, or warranty, request is made over the Internet to a web site that provides the data input and output ports.

9. A computerized valuation exchange according to any preceding claim, wherein the request includes the sale price or the amount of the loan, or merely the identity, such as the address, of the property.

A computerized valuation exchange according to any preceding claim, wherein the stored data includes the dates and values of any sales that have occurred, grouped by geographical area, for as long as sales records have been kept.

11. A computerized valuation exchange according to any preceding claim, wherein the determination of property values takes place on a regularly recurring basis, after new sales data is received, and requires no physical inspection of the property.

12. A computerized valuation exchange according to claim 7, wherein the calculation of a valuation, or a warranty fee, takes place immediately in response to receipt of a warranty request.

13. A method of operating a computerized valuation exchange to receive valuation requests in relation to properties in a region, and to provide a valuation advice in response to each request, the method comprising the steps of: receiving data about property sales within the region, including the date and value of property sales that have occurred in pre-determined geographical areas of the region; calculating statistics relating to the property sales data in each geographical area; deriving capital growth indices for each area based on the change in sale price of properties that are sold more than once; deriving residuals to describe the distributions of deviations from the capital growth indices; deriving the distribution of errors between purchase prices and a range of possible purchase prices using the residuals; estimating a range of property values for each property within the region using the statistics, indices, residuals and errors; storing data representing the determined range of property values; receiving a valuation request at a data entry port; calculating a value automatically in response to a request, using the range of property values; and, outputting, at a data output port, the valuation and the warranty.

14. A method according to claim 13, wherein the step of receiving a valuation request includes receiving a valuation request at the time of a purchase, and the valuation request includes an advised value for the respective property, and the estimating step takes account of the advised value when estimating the value.

A method according to claim 13, which also takes account of one or more of the following factors: historic sales data for that property, the appropriate capital growth index and the residual; previously determined property values in the immediate vicinity of the property.

16. A method according to claim 13, wherein the method uses a combination of the factors to estimate the range of property values.

17. A method according to any one of claims 13 to 16, wherein a value from the estimated range is taken to give a single determined property value.

18. A method according to claim 13, where 17, where the value is median.

19. A method according to any one of claims 13 to 18, wherein the valuation is used to provide a guarantee (or warranty) of a value for the property at the time of the loan.

A method according to any one of claim 19, wherein the valuation, or warranty, request is made over the Internet to a web site.

21. A method according to any one of claims 13 to 20, wherein the request includes the sale price or the amount of the loan, or merely the identity, such as the address, of the property.

22. A method according to any one of claims 13 to 21, wherein the stored data includes the dates and values of any sales that have occurred, grouped by geographical area, for as long as sales records have been kept.

23. A method according to any one of claims 13 to 22, wherein the determination of property values takes place on a regularly recurring basis, after new sales data is received, and requires no physical inspection of the property.

24. A method according to claim 19, wherein the calculation of a valuation, or a warranty fee, takes place immediately in response to receipt of a warranty request.

A computerised valuation exchange substantially as herein described and with reference to the accompanying drawings. 17

26. A method of operating a computerized valuation exchange substantially as herein described and with reference to the accompanying drawings. Dated this twen'ty-third day of April 2003 Residex Pty Limited Patent Attorneys for the Applicant: F B RICE CO