First, note that the time adjustment is core to market analysis.  It is not about “comparing comps.”  It quantifies two things. The change in:  1) the value of the dollar and, 2)  market segment price levels.

Reference:  FHFA Working Paper 24-07:  Disparities and Time Adjustments, the paper by William Doerner and Scott Susin (FHFA) relates racial bias to time adjustments! 

What to Do?

Three methods are suggested as being in “common” use by appraisers.  But recall that, at most times, most appraisals evade trend analysis by simply putting in zero  or NA – a “level” market segment.

First method – “grouped” data, using annual or quarterly “time groups,” broken down into monthly averages.  Problem:  grouping pretends all sales take place at the group mid-point date.

Unfortunately, this means that two identical sales, next door to each other, selling one day apart, would have to be adjusted for the change between group-monthly averages, or groupedquarter averages (or medians).  In statistics or econometrics, this is called data discarding.  If you have the exact date, use it.  Don’t “average” the date to the time-middle of the group.

Second method – “paired sales” (“repeat sales” in AVM lingo).  The theory is that a house sells twice at just the right times, magically calculates the market change in price.  It does not.  Problem:  Most time adjustments are needed for less than a 12-month back time span.

The likelihood that both sales are helpful dates, and of “normal” motivation, and in unchanged features, condition, and equal age – is dubious.  Most will be “fixers,” or sellers motivated in non-typical ways.  Any resulting time adjustment would be confounded with these non-time related changes.

Third method– “visual plot regression” means graphing competitive sales with time (sale date) on the x-axis, and $ price on the vertical y-axis.  Problem – a straight-line regression doesn’t recognize any trend-change within the visited time span.

However, changes in trend can be seen visually.  (Twelve months or more works well.)

A straight-line simple regression provides the daily change, and is applied to the number of days-back for each comparable.  Fitting a “third-order” polynomial helps point out any trend change, which can be refined to a “spline-linear” or even directly off the curved polynomial.  Outliers immediately show.

The Answer – The best result has four requirements:

1)  The right data selection – competitive market elements;

2)  The right timespan – to include the oldest comparable,

2)   Resolve any outliers  – identify, decide, and explain;

4)   Handle trend changes –  identify, and present.

Simple!  Time-series is unique, but easy to understand and use.

The analysis software is free or inexpensive.  You can use spreadsheets, R (RStudio), or even word-created AI algorithms!  What is important is Your understanding of how to see, how to create, and how to present graphs and statistics into 1004 forms and the new interactive forms.

All known current residential forms software easily accommodate the days/dollars adjustment.

Summary — The Grouped method forces a categorical variable onto measurable data.  Paired sales provides non-matching time spans.  This method is not useful in practical practice.  Both methods were developed as theory in the olden-days of hand-collected data.  Both discard/ignore available market data.

Contrarily, visual regression is a modern, data-driven method, suitable to today’s computer power and easy market visualization.  It entails rigorous time-series methods in econometric analysis.

The FHFA report bias conclusion exposes slapdash analysis, missing enforcement, and missing education.

At Valuemetrics.info, we teach the recommended “Market Price Indexing” (MPI)© method.  It forms the basis of Evidence Based Valuation (EBV)©.  The Stats, Graphs, and Data Science classes, free webinars, and free on-line videos are available to any appraiser (even non-residential types).

Visit GeorgeDell.com, or Valuemetrics.info now.