Monday, September 29, 2014

Homequant.com Introduces REO and Foreclosure AVM

PRESS RELEASE

http://www.pr.com/press-release/584131

New York, NY, September 29, 2014 --(PR.com)-- Prime foreclosures and charge-offs are expected to stay elevated due to the continued squeeze in the Jumbo mortgage market, high incidence of short sales and the highly probable tsunami of HELOC defaults hitting the market in early 2015. If the HELOC hypothesis comes to pass, a new generation of REO and foreclosure (“Foreclosure”) AVMs geared exclusively towards that segment would be mandatory.

In fact, everyone from the traditional AVM houses to the listing services to the national brokerage houses has realized that the foreclosure markets are not short-lived or temporary. Actually, the overall housing market has become semi-permanently bimodal (primary and foreclosure), requiring significant back-to-the-drawing-board valuation re-engineering.

Under the traditional AVM development process only the recent arms-length sales (often aided by the discounted seasoned listings to simulate the most recent market) are used to create representative sales samples to develop Multiple Regression Analysis (MRA) models, which are then applied on to the populations the samples are derived from. In other words, the traditional modeling samples ignore all foreclosure and short sales.

However, to develop Foreclosure AVMs, the experts at Homequant derive modeling samples from the foreclosure-related universe only, to avoid having to distort the final values by applying some heuristic discounting factors. When the AVMs are developed as such, the final values are more in line with that segment of the market, addressing especially the sub-markets which inherently deviate from the median market.

Of course, to bolster the sample size, they often group and model multiple contiguous markets together; for example, if the local MLS covers three counties, they tend to model them together drawing all of their foreclosure and short sales into the mix. Obviously, it’s easier for them to generate sales samples in those Metropolitan Statistical Areas (MSA’s) where foreclosures are common or are elevated.

The lenders holding large inventories of REO and foreclosures (late stage) would also be better served with the proposed Foreclosure AVMs, not only to price their portfolios more accurately, but to prevent appraisal frauds as well. Those AVM values should be used as control values to trigger the Supervisory QC.

If you’d like more information about their Custom AVM solutions, or to schedule an interview with them, please email them at: contact@homequant.com

About Homequant
Homequant is the inventor of the comparables-based valuation of simulated subjects. The President of Homequant recently explained their invention: “There are roughly 90 million single family homes in the US and, on average, 3% of that universe annually sells. By inventing the concept of the simulated subject, we are able to value those 97% unsold properties by storing only the 3% sold data. The home valuation industry will soon recognize the significance of our invention.”

Homequant’s President published two books on econometric AVM.

Contact Information
Homequant, Inc.
Sid Som
718-314-4081
www.homequant.com

Sunday, September 28, 2014

Books on Automated Valuation Modeling (AVM)

(Click on the image to enlarge)
According to Amazon, "Anybody can read Kindle books—even without a Kindle device—with the FREE Kindle app for smartphones, tablets and computers."


REO and Foreclosure AVM – The Homequant Way

http://www.prlog.org/12376570-reo-and-foreclosure-avm-the-homequant-way.html 

Most banks and mortgage houses have been buying AVM values from the leading vendors to cater to their needs in the front-end and mid-end. However, since the recent housing bust, their back-end (collections, foreclosures, short sale, REO sales, etc.) needs have exploded. Obviously, the AVM values that are meaningful for the front-end and mid-end are practically useless in the back-end.

Here are some of the reasons why specialized REO and foreclosure AVMs ("Foreclosure AVM") would be needed to address the fast-growing needs of this market segment:

1.  An Overall Discounting Factor is Only a Surface Correction

Let’s say, in major market X, the market differential between the primary market and the foreclosure market is 30%. On the heels of this market statistic, if a bank buys current AVM values from a vendor and tries to fit a 30% haircut to its foreclosure portfolio, it would make a serious mistake by only emphasizing surface corrections, thereby distorting the sub-markets (a.k.a. pockets) that tend to deviate from the norm.

Even a local housing market consists of many diverse sub-markets that tend to be econometrically different from the smooth median corridor. Therefore, those deviating sub-markets would be grossly mispriced should a generic haircut is applied; in fact, some would be grossly overvalued while others would be significantly undervalued, thereby lowering the market reliability of the entire portfolio.

2.  Foreclosures are Often Disproportionately Clustered in Sub-markets

Since foreclosures are often disproportionately higher or clustered in certain sub-markets, using generally discounted AVM values as described above would be irrational, particularly when large portfolios are negotiated, causing more trouble for the retail brokers and small homebuilders who would try to rationally work through their inventories.

Of course, today, foreclosures are more wide-spread, extending from the prior sub-prime and Alt-A into the prime portfolios. Therefore, they require more specialized valuation, including Foreclosure AVMs. The high-end trophy properties should not be subjected to any AVM’s, instead professionally hand-worked.

3.  Foreclosures will Plague Prime Portfolios for Several More Years

Prime foreclosures and charge-offs are expected to stay elevated due to the continued high incidence of short sales and the highly probable tsunami of HELOC defaults hitting the market in early 2015. If the HELOC hypothesis comes to pass, a new generation of Foreclosure AVMs geared exclusively towards that segment would be mandatory.

In fact, everyone from the traditional AVM houses to the listing services to the national brokerage houses has realized that the foreclosure markets are not short-lived or temporary. Actually, the overall housing market has become semi-permanently bimodal (primary and foreclosure), requiring significant back-to-the-drawing-board valuation re-engineering.


The Practical Aspect
Under the traditional AVM development process only the recent arms-length sales (often aided by the discounted seasoned listings to simulate the most recent market) are used to create representative sales samples to develop Multiple Regression Analysis (MRA) models and are then applied on to the populations the samples are derived from. In other words, the traditional modeling samples ignore all foreclosure and short sales.

However, to develop Foreclosure AVMs, the experts at Homequant derive modeling samples from the foreclosure-related universe only, to avoid having to distort the final values by applying some heuristic discounting factors. When the AVMs are developed as such, the final values are more in line with that segment of the market, addressing especially the sub-markets which inherently deviate from the median market.

To bolster the sample size, they often group and model multiple contiguous markets together; for example, if the local MLS covers three counties, they tend to model them together drawing all of their foreclosure and short sales into the mix. Obviously, it’s always easier to generate sales samples in those Metropolitan Statistical Areas where foreclosures are more common.

Traditional AVMs require significant amount of GIS and spatial analysis to define and quantify non-fixed planes and pocket areas. Those variables are then added to the MRA model, along with the fixed neighborhood variables. Since foreclosures are more clustered and/or pocket-oriented, the concept of non-fixed planes does not take form.

Therefore, instead of the traditional three-stage MRA model – additive, multiplicative/log-linear and GIS – they use a two-stage model structure for Foreclosure AVMs.

IWhen the foreclosure pockets within the model could be clearly identified and defined (using Latitude & Longitude), they are then introduced in the equation as linearized clusters or pockets. Any more GIS/Spatial analysis tends to be overkill.

The generally accepted quality metrics like Coefficient of Dispersion (COD), Price-related Differentials (PRD) are unnecessary. Instead, they encourage their clients to review the output quality by forming a valuation review group consisting of internal foreclosure experts, selling brokers and AMCs (when involved).

In any case, the objective of the Foreclosure AVM is to manage and mitigate losses so these AVM values do not have to be as surgical as the traditional AVM's. When those Foreclosure AVM values are compared to their traditional counterparts, they will be significantly lower in a market-meaningful manner.

The lenders holding large inventories of REO and foreclosures (late stage) would also be better served with the proposed Foreclosure AVMs, not only to price their portfolios more accurately, but to prevent appraisal frauds as well. Those AVM values should be used as control values to trigger the Supervisory QC.

Simply put, the incoming REO appraisals must be compared against the AVM values and those that deviate from the internally acceptable range (say, +/- 20%) must be flagged for the Supervisory QC. Without such internal controls, those portfolios would be susceptible to appraisal frauds. AMCs should also use a similar approach to score and rank their appraisers.

Wednesday, September 24, 2014

Why Homequant.com is the most accurate free home valuation system today!

Traditionally, appraisers use comparable sales to value actual subject properties (having real house addresses with real property characteristics). The high costs of maintaining the subject population often gets reflected in less-than-adequate maintenance of the ever-changing sales population, thus producing obsolete comparable sales ("comps") and, in turn, outdated subject valuations.

Our solution demonstrates that a Defined (a.k.a. Simulated) subject ("subject") serves as good a purpose as an actual subject property, freeing up the hefty costs of maintaining the subject population. Our system, www.Homequant.com, - built on our solution - is therefore predicated upon empowering our users with a sophisticated yet easy-to-use methodology to define subjects, an up-to-date sales database, and an advanced comps matching and scoring tool so that the users can meaningfully undertake a series of experiments on-the-fly to arrive at their own value conclusions. In fact, in the process of defining their subjects, our users are essentially researching and duplicating actual subjects in their experimentation, without blindly relying on the often-outdated public record.


Additionally, unlike an institutional client, an average home buyer – in Free Home Valuation industry where we are positioned – is generally interested in valuing a handful of subject properties so researching and capturing such data tend to further motivate our users. 


Unquestionably, given the anemic health of most taxing jurisdictions nowadays and the falling tax base, subject populations are nowhere as up-to-date as they used to be in earlier years. In fact, the lack of field inspections by the assessing staff results in significant delays in the capture and inputting of the physical changes in the assessment system. Therefore, when a potential buyer or an investor depends on such subject data, the resulting valuations are generally less-than-perfect, to say the least. Alternatively, when a targeted subject is individually researched, inspected and manually entered – as is the case in our system – the resulting valuation becomes significantly more reliable.

The housing population consists of the (unsold) subject population and the sold population. Since only the most recent (one to two years) arms-length sales could be used as comps to value subjects, the general split between the subject population and the comp population is 95% and 5%. Considering our system accurately functions off 5% of the population, our system does not succumb to the data logistics, nor does it require a fortune in terms of investment in maintaining the data, allowing us to remain focused on updating the sales data and enhancing the technology to keep our visitors enticed.

Homequant system consists of two separate yet complementary modules: Valuation (micro) and Sales Query (macro).


Valuation Module

We are aware that sales by themselves are not comps; they have to be equalized and converged to the subject to become true comps, so the Homequant system has been built to provide for an excellent adjustment mechanism that users can experiment with and generate the optimal comps. In the process, we adhered to the strict market economics by attaching the highest importance to the location, followed by the property characteristics.


Defining Subject

In terms of the location, it is top-down, meaning users have to first select the county, then the city/town and the zip code and finally the actual address. For example, if the user is valuing a subject located at 46 John Street in Forest Hills, Queens, New York, s/he has to select Queens, Forest Hills, and Zip code (11374 or 11375) and then enter the street address, thus prompting comps to align with that location vertical. As the actual address is entered, a ‘validate address’ button below will pop up. Once the address is entered and the validation button is pressed, the address will be internally validated against the Google Earth database. If Google Earth finds a match, a new pop-up will confirm, absent which the address requires rework.

In terms of the property characteristics, Homequant uses the three most important continuous variables, namely land size, building size, and age (derived off of year built). Therefore, Homequant requires users to enter these three values. The subject definition is complete once the location and these three property characteristics values are entered.


Comps Selection

Of course, the most important exercise in Valuation is to select the right pool of comps. Once that’s done, Homequant adjusts, equalizes, ranks and finally returns the five best comps. In the process, Homequant requires users to select an appropriate comps selection range, between 10% and 40%, in order that an internal matching algorithm can pool and return the best five comps from the defined range. If a user selects a 10% comps selection range, comps within that land size, building size and age range of the defined subject would be evaluated for pooling. For instance, if the subject has a land size of 4,000 SF, building size of 1,500 SF and age of 50, comps between 4,400 and 3,600 SF in land sizes, 1350 and 1650 SF in building sizes and ages 45 to 55 would qualify for pooling. Needless to say, only those comps that pre-satisfy the location criteria would qualify for pooling as well.


Distance Radius

In addition to selecting the comps from a defined range, they must come from the vicinity within the subject’s location vertical. In other words, all factors remaining equal, comps that are closer to the subject are the better comps. In line with this established market theory, Homequant provides a ‘distance’ option allowing users to restrict the inflow of the comps between one and five miles. Alternatively, in rural counties where comps could be far apart, this option might be easily bypassed by selecting ‘none.’ Either way, this is a critical option in comps selection, without which the comps closer to the subject could be overridden by the distant ones.  


Comps Dollar Adjustment

In certain illiquid housing markets – for example, million-dollar-plus and waterfront – comps are often few and far between, forcing a much wider selection range – say within 30% of the subject’s. In those circumstances, comps require dollar adjustments. Homequant allows separate adjustment coefficients for land, building and age. If the subject sits on a 10,000 SF lot and the first two entering comps carry 8,000 and 12,000 SF respectively, then the first comp would be adjusted up (+) 2,000 SF, while the second would be adjusted down (-) 2,000 SF. If a land adjustment coefficient of $10 is chosen, $20,000 would be added to the sale price of the first comp, while $20,000 would be subtracted from the sale price of the second comp. This adjustment process would repeat for the building size as well as age.Without the proper selection range backed by the proper dollar adjustment, sales do not necessarily become comps to value a subject. Homequant’s competition in the free home valuation industry fails to recognize that and provides no such mechanism, thus forcing users to settle for eyeballing comps, coupled with unsound quantitative judgment.


Time Adjustment

Housing market is very dynamic, causing prices to be volatile – prices take off or swoon within a very short period of time, so two similar home sales nine months apart are not the same; they require sound quantitative time adjustment. An educated customer knows the dynamics in the market, meaning if the prices are on upward or downward trajectory. If the prices moved up 10% over the past one year, the older sales must be adjusted ‘up’ for time. Likewise, if the prices declined 10% year-over-year, older sales must be adjusted ‘down’ to be usable as comps. Homequant allows both positive and negative time adjustment to cure sales for time.


Valuation Date

Many valuation professionals – money managers, appraisers, assessors, etc. – need to value subjects for a different valuation date, other than the current period. Assessors often have a backward taxable status date; money managers often reprice a portfolio for a backward date; economic analysts often need to revalue properties for a forward date. Homequant allows valuation for all three valuation dates – current, forward and backward. On 11/1/2013, if a user wants to value a subject for 12/31/2013 projecting 1% monthly growth in prices, Homequant will automatically calculate the time-adjusted value as of that date – no additional entry or gyration will be needed. For the current and forward dates, all sales will be adjusted up, while for the backward date, all newer sales beyond the valuation date will be adjusted back. Again, Homequant’s competition in the free home valuation space does not offer any mechanism to adjust for time, let alone allowing flexible valuation dates.


Ranking Method

Once the comps are pooled, adjusted and scored they need to be ranked. Homequant provides three different methods: distance, sale date and least adjustment. When ‘distance’ is chosen, the qualified comp which is physically nearest to the subject is comp 1. If the ‘sale date’ method is chosen, the qualified comp with sale date closest to the valuation date becomes comp 1. Under the ‘least adjustment’ ranking method, sum of all dollar adjustments, including time adjustment, for each comp is compared and the one with least adjustment becomes comp 1. In evaluating the least adjustment amount, +/- signs are ignored. In other words, a total adjustment of $1,000 will supersede -$6,000.  While the distance and sale date methods are meaningful for the non-professional users, the least adjustment method would be a powerful tool for the professionals and advanced users.      


Minimum Comp Requirement

Homequant requires a minimum of five comps to value a subject so users would probably start at the bottom of the selection range, i.e., 10% and gradually work up depending on the results obtained. If less than five comps are pooled, a slider is provided to adjust the curve in increments of 5%, just like in the manual input box, to avoid having to manually enter into the boxes. In order to allow more qualified comps to come in, the distance could be increased, say from one mile to two miles, in combination with the widening of the selection range. For example, if the initial selections range of 10% across land, building and age and distance of one mile returned only two comps, the range could be increased to 20% and the distance to two miles.


Quantitative Output

Once the five comps are accepted a comps grid with details of each comp as well as the adjustments would be produced. Specifically, the grid would show the exact location including the street address of each comp, physical attributes in terms of land size, building size and age, sale price and sale date and the overall adjustments made to sale price to generate the adjusted sale price which, in turn, becomes the comparable metric, after having neutralized the physical and market differences in them. In order to study the output multi-dimensionally, the output table may be sorted by any of the columns listed there by simply clicking on the column header. This sorting is restricted to this output table without affecting the ranking method which controls the final comps grid. 
  

Spatial Output

All sales are geo-coded with Latitude and Longitude coordinates before uploading on to Homequant, paving the way for precise spatial views as well as concise spatial outputs. The final five comps – and the defined subject – are simultaneously shown on a Google Map, allowing users to visualize - and swap - their prior selection. In fact, this gives them a renewed chance to revise and anchor their final five, before moving on to valuing a different subject altogether. Alternatively, this spatial view also helps them understand why the fifth comp, for instance, had sold for 30% higher than their peers (perhaps, close to a lake). An educated user would instantly disregard that comp from the line-up, despite having met the basic size, distance and location criteria. Since Google Street View is also integrated on the same output, they can also view the actual homes as well as their surroundings.  It is very useful for the remote buyers researching into the area.


Comps Grid and Subject Value

The comps grid - which follows the selection of the final five comps - would be ranked by the ranking method chosen by the user. Simply put, if distance is the chosen as the ranking method, the comp that is physically nearest to the subject would show up in the grid lineup as comp 1. If no ranking method is chosen, the default method - Sale Date - would rank the grid.

In addition to the grid, a complete percentile distribution analysis of the full spectrum of the Adjusted Sales Price (ASP) has been provided right underneath the grid, thus translating the isolated sale prices into a market snapshot of where the probable values could be for the various interest groups. For example, an aggressive investor might consider a range between the 5th and 25th percentile, while an informed home-buyer would probably consider the 25th to 75th percentile as the most probable range, with the median (50th percentile) ASP being the most probable indicator of the market price, singularly, and so forth. 

Again, Homequant’s competition in the free home valuation space does not offer any mechanism to tabulate the selected comps on a grid or to statistically analyze them collectively.


Sales Query Module

In addition to valuing individual subjects, Homequant also offers a macro choice, called Sales Query, to help users understand the broader market. Users can learn a great deal of a specific market or a group of markets in many different permutations by spending a few moments here. In fact, the Valuation and Sales Query modules are designed to complement each other, without any unnecessary duplication of themes. Case in point: When a user is uncertain of the value parameters of a set of comps, s/he can visit this macro module to see how the results from the comps compare with the universe they are drawn from. If the user wants to learn more about the Forest Hills (NY/Queens) housing market before toying with individual subjects, s/he would go to Sales Query and run the global stats there.

Since sales as old as one year ago are included in the sales database the same time adjustment and valuation date options are incorporated in Sales Query. Again, in valuation date, the default is the current date so all sales would be time adjusted to this date (assuming a time adjustment factor is entered). Likewise, a forward or a backward date could be experimented with. When a time adjustment is sought, the output box would show different values between Sale Price and Adjusted Sale Price. If no time adjustment is sought, they would be identical. In Valuation module, the Adjusted Sale Price comprises of the sum of the feature-wise dollar adjustments and time adjustment, while in Sales Query, Adjusted Sale Price includes only the time adjustment component considering there are subjects here.

Sales Query output is simultaneously displayed on the Google Map as well, allowing new users to see where the selected node is located within the county.

To summarize the merits of our solution – the Homequant free home valuation system: 
  • Homequant solution is predicated upon the premise that user defined and researched subjects are as good, if not often better than the subject data from the public record which eliminates the need to maintain the gigantic subject population. It also translates to a significant saving in working capital that is otherwise better utilized in maintaining an updated sales population, resulting in more up-to-date sales and thus more accurate values.
  • Homequant solution – combining user-defined subjects, an up-to-date sales database, and an advanced self-directed tool (subject definition and comps matching, adjusting and ranking) – optimally empowers users, allowing them to arrive at their own value conclusions, objectively and iteratively – a far better approach than offering some model-driven frozen values that no one could decode.
  • Homequant recognizes that selecting comps from a given location or strata is not enough so it allows dollar adjustments to the size of the land and building, as well as to account for the annual depreciation of the building.
  • Homequant recognizes the ever-increasing volatility in the market – rapid rise and fall in prices – and therefore allows an effective time adjustment mechanism.
  • Homequant recognizes its audience comprises of professional and non-professional users so it provides for variable – current, forward and backward – valuation dates.
  • Homequant recognizes that the traditional ranking of comps by the distance alone is inadequate for the advanced users, so it provides for additional ranking methods like least adjustment and sale date.
  • Once the final five comps are scored, ranked and accepted, they are tabulated on a traditional appraisal grid, allowing professional users to relate to their comfort zone.
  • In addition to traditional value parameters like average and median price of comps, Homequant offers a full percentile distribution curve to accommodate advanced and professional users as well. Many advanced users like money managers, loan officers, portfolio analysts, etc. appreciate the addition of the percentile distribution.

While Homequant is not a full-blown property valuation system, this is however the most advanced valuation system available in the ‘Free Home Valuation’ space today to:

1. educate novice users to understand the local housing market more objectively,

2. offer a good bottom-up second opinion to the non-professional yet well-informed users,

3. help advanced retail users like home sales and mortgage professionals get a quick preview of a subject valuation, from anywhere,

4. help sophisticated wholesale users like real estate analysts and portfolio managers quickly sample the value of a given portfolio,

5. help assessment review officers and appeals/hearing courts determine if full-blown professional appraisals would be required.

So, how does Homequant's "Defined Subject" concept help the Valuation Community at large?

Often, the Valuation Entrepreneurs fail to make inroads into the Valuation business due to the high cost of the unsold  population data which tends to be about 95% of the entire population. Because of Homequant's discovery of the "Defined Subject" - meaning user-defined subjects - those entrepreneurs can jump into the business now, knowing full well that the sold data alone - generally available free of cost or for a token price from the taxing jurisdictions - would do the trick, without having to invest a fortune in acquiring and warehousing the unsold data which, in terms of quality, is questionable at best, to begin with.


The President of Homequant recently explained our invention: “There are roughly 90 million single family homes in the US and, on average, 5% of that universe annually sells. By inventing the concept of the user-defined subjects, we are able to value those 95% unsold properties by storing only the 5% sold data. The home valuation industry will soon recognize the significance of our invention.”

In a nutshell, the Homequant solution provides a very realistic – yet advanced – solution to the pros and non-pros alike by perfectly bridging the gap between the pseudo free home valuation sites showing only a list of comps and the enormously expensive professional appraisal systems.  

Try it. You'll love it! 

Here is the site address:

Thanks,

Sid Som MBA, MIM
President
Homequant, Inc.

contact@homequant.com

Additional reading:
How to value a subject on Homequant site (step-by-step):
http://homequant.blogspot.com/2014/06/how-to-value-subject-property-on.html 

Friday, September 19, 2014

How to Evaluate an AVM Vendor



Since AVM has now been around for a while, the vendor market is predictably quite crowded too. Those vendors not only cater to the mortgage industry (origination to portfolio management to collections and foreclosures) but they also supply their values to a host of other financial servicers, as well as to a growing list of public agencies including the department of assessments, assessment reviews, state equalization boards, and small court systems.

Given the growing need for the AVM values and a crowded landscape of their providers, it is critical to understand how to evaluate them before entering into a lengthy contract. While AVM values could be wide-ranging, for the purpose of this book, focus will be on the mass-modeled values. Additionally, the vendors who openly resell a mix of 3rd party values (without the hide-and-seek of the private labeling) to compliment their other product lines are also outside the scope of this book. Here are the points of safeguards one must consider:


1. In-house Value vs. Private labeled Value

 Other than the exclusive resellers, the vending crowd can be broadly grouped into two categories. The In-house value vendors have their own in-house or contract researchers developing and updating the AVM values, while the Private-labeled folks tend to outsource the development to other research houses and consultants with a no-compete, often on a revenue-sharing basis.

During the prior (2003 through 2007) real estate/mortgage boom, many low-cost research houses sprang up in India and other Southeast Asian countries to take advantage of the local high-end yet relatively inexpensive quantitative skills. Naturally, the In-house folks, from the very beginning of the sales conversation, would be more forthcoming and transparent than their counterparts. 

For example, the In-house folks would be more than willing to work with custom samples (i.e., to populate samples provided by the potential buyers) considering their inherent (Valuation and IT) strength to outpace the competition, while the Private-labeled folks would invariably point almost everyone to some frozen (web-based) values. Therefore, depending on the potential size of the contract, one should ask to populate a custom sample (better yet, a series of samples as sales dialogues mature) where recent appraisals/work-ups are available so the overall deviations (COV, COD, etc.) may easily be computed to establish the effectiveness of the vendor’s values. Once the top choices are zeroed in, paying a visit to their facilities is advisable. This two-step selection process will help avoid a lot of agony down the road.


2. Quality of Values

To differentiate the qualified from the unqualified, it is always a good idea to ask if they are S&P or OCC (or other well-known bodies) compliant. The vendors that are compliant know very well that their long-term existence in the industry is predicated upon the quality of value – industry standard compliance is the gold standard. The unqualified population, on the other hand, will offer a million excuses as to why that is totally unneeded. Here are some warning signs: ‘it is too time-consuming’; ‘we revalue all properties nightly’; ‘our logistics don’t allow that – we work with multiple data vendors’; ‘we don’t believe in those old compliance standards – we have advanced in-house QC measures in place,’ etc.

Of course, the public agencies are often bound by the International Association of Assessing Officers’ (IAAO) guidelines and standards (Coefficient of Dispersion, Price Related Differential, etc.), so the proof (from those who also cater to the public agencies) that they meet and exceed those requirements should be requested. Warning signs: ‘what is IAAO?’; ‘we are OCC-compliant so we don’t need to meet any other requirements and standards,’ etc. Public agencies should specifically look for values that are always IAAO-compliant.

Of course, this safeguard does not apply to the fast emerging group of vendors that specializes in custom modeling and value generation using client’s own internal data, meaning (private) portfolio data or (public) assessment roll data.


3. Nature and Frequency of Value Updates

The frequency requirements vary. In order to work with the mortgage originations, vendors need to frequently rerun their models (with more recent sales) to stay close to the current market conditions, whereas portfolio managers tend to be more P/L-oriented (quarterly to annual), and public agencies need conform to the taxable status dates (annual).

While some regional/national data warehouses also produce their own AVMs, most AVM vendors either buy from or form a Joint Venture with the data warehouses to receive the data, with monthly updates, considering the underlying AVM data – population and sales – are quite expensive.

In a volatile market such as this, even the best engineered models require frequent updating and fine-tuning (preferably monthly - at least quarterly), so it is good to understand the nature and frequency of their value updates. Many AVM houses tend to rerun their models annually, but time-adjust in between, which may not be good enough. Those models must be updated at least quarterly.

The best way to find out how frequently and effectively (whether they are rerunning the models or just time-adjusting the old model values) they are updating their values is to ask them to repopulate the same custom sample(s) in the following month(s) as sales conversations continue. One can thus easily determine how up-to-date their values are. If they return the same set of values (I doubt their salespeople would even allow that!), they should be avoided.

If one can identify a time adjustment pattern (say, adjustments by specific geographies like city or county) in vendors’ values, they should be considered with a grain a salt as well. If their new values are different yet as surgical in line with the recent dynamics of the market (the way a multiple regression equation with 50+ independent variables is supposed to work!), one can infer that they truly reran the model with more recent sales, and therefore their AVM values would be more effective and thus would make a significant difference for the user organization.


4. Encourage Competition

 Instead of using one vendor, it’s good to try two vendors (separate them along the portfolio/geographic lines) so they would always know that there is competition even within the client organization and that they have to continually prove their worth to keep the account.

It is equally important for the public agencies; for example, if the department of assessment uses one vendor (to produce the roll or to validate the roll), the assessment review department must use an alternate vendor to avoid having to justify the ‘double dipping’ to the legislators.
Internal competition will invariably force the vendors to intensify their R&D, knowing very well that losing a key client could be the kiss of death meaning others often follow suit. In other words, when the word spreads about vendor X losing a key client, other existing clients become more cautious, often aggressively without proper investigation.


5. Buy According to Your Needs

AVM values can be quite expensive, so it is essential to perform a needs analysis internally before putting out a RFP or contacting a group of vendors. For a portfolio manager needing to revalue portfolios quarterly or annually should always sign contracts keeping that frequency in mind, thereby paving the way for some substantial savings. Alternatively, while negotiating for the AVM values for in-house portfolios, it is wise to try to encourage the vendor to run the complementary static pools or vintage and migration analyses for free or at a significantly discounted rate.

The declining demand for mortgage products since the last housing meltdown has lowered the demand for AVM values, which in turn has hurt the industry, particularly those in the the outsourcing industry (Knowledge Process Outsourcing or KPO’s). Therefore, the AVM vendors – in pricing their products – ought to be more flexible today. Won’t hurt to ask!

Note - This book is available on Kindle (Amazon - query 'Sid Som').

Thanks,

Sid Som MBA, MIM
President
Homequant, Inc.
contact@homequant.com