Time trending is a way of adjusting sales prices to account for changes in market conditions since the sales occurred. The end result of the process is a set of adjusted sales prices that have all been trended to the same point in time. This point in time is what we call our appraisal date, and it is always June 30th of the year preceding the reappraisal year. Once the sales have been adjusted in this way, we can use them in our mass appraisal modelling to determine the most likely market value of other similar properties on the same appraisal date.
To perform the time trend analysis we start by calculating a sales ratio for each sale in our data set. The sales ratio is the sale price divided by the actual value that we have placed on the same property from our previous reappraisal cycle. If the resulting ratio is greater than 1, the property sold for more than our valuation; if it is less than 1, the property sold for less than our valuation.
For sales that occurred close to our prior reappraisal date, one would expect the sales ratios to average about 1. Since market conditions often change over time, it is typical to see average sales ratios diverge from 1 before and after the prior reappraisal date. We measure this by creating a graph of sales ratios (on the vertical axis) against time (on the horizontal axis). To illustrate this concept, refer to the following time trend analysis graph for vacant land in economic area 6. On this graph, the horizontal axis is numbered in months since June 30th, 2007: the previous reappraisal date is therefore month 36 (June 30th, 2010) and the current reappraisal date is month 60 (June 30th, 2012).
The Gunnison County Assessor's Office uses a statistical software tool called SPSS to graph and analyze time trends. One of the functions in this tool is the ability to draw a time trend line on top of the graph. This line is called a Loess line; SPSS generates this line by performing what is known as locally weighted regression analysis. Put simply, this is a way of drawing best fit lines through the data points in situations where there may be several different trends going on at different times. As indicated on the graph, market values dropped about 1.5% per month between July 2007 and March 2011 (months 1 through 45), and thereafter stayed approximately static through June 2012.
The next step for us is to use this time trend to adjust all our sales to the new appraisal date. For example, if a sale occurred in April 2010 (month 34 on our graph) we would adjust the sale price down by 16.5% (11 months times 1.5%). If a sale occurred in October 2011 (month 52) there would be no time adjustment and the adjusted sale price would equal the actual sale price.
This approach to time trending is a standard appraisal practice used throughout the world, and is recommended by the International Association of Assessment Officers.