We hear noise in appraisal work on trend analysis, time adjustment, and “market conditions.”
These overlapping terms are related but often mixed in their use. Lack of clarity/definition is perplexing! The confusion may be explained by:
- Lender/investor goals may vary.
- Modeling differences in “signal vs. noise.”
- The relevance of seasonality depends on intended use.
First, the goal. Most commonly, the appraiser needs to make adjustments for time (“changed market conditions”) over the period of time for each comparable sale.
A client-user may also want to know the trend for a given time period, say one year or three years, which may not coincide with the most dated comparable sale date.
And other users may ask for a probable sale price sometime in the future, after a given or estimated market exposure time after the current date, such as for relocation companies, and construction.
The signal is the trend, or the price index needed for either the time adjustment, or the longer trend span time. The algorithmic goal is to “smooth” to the trend (commonly least-squares), while avoiding individual sale variation (noise from other property features). This works after the data selection process to moderate (control for) to relevant data – by other features, such as living area or site size.
Seasonal trends are regular in nature, and may be a noise, or a signal, depending on the analytic goal.
(The use of the words “to moderate” recognizes that control to exact similarity is not possible. But it does reduce the influence of non-competitive sales and data outliers.)
Seasonality plays into certain market segments – such as ocean-side summer vacation, or ski-resort areas. In such cases, some users (like long-term investors) will want to see the trend over a longer period, probably beyond the seasonal variation. from say summer to winter . . . Two different strategies, requiring different analysis.
For the appraiser, working to a time adjustment from a 7-month-old sale, seasonality smoothing is “ignored,” in that a proper time-series analysis should account market conditions change.
For the longer-term trend analysis, it is likely the user will want to “even out” or moderate the effect of seasonal demand. The search for fundamental value or intrinsic value or forecast value is different.
Time-adjustment is different from trend analysis! The goal is different.
Analytically, the time-of-year itself becomes a predictor (adjustment) variable. The recognized method here is to go back several years, then apply the time-of-year adjustment to the trend. Also, this would require the user/client to identify (within the appraiser’s scope of work) the time-of-year-value desired. The appraiser should not assume a mid-range (autumn or spring) value.
To sum up: an appraiser should not separately time-adjust for season. It should be embedded in the controlled/moderated competitive market data set.
This explanation does not resolve the conflation of the different but common terms used in current appraisal practice. The influence of dominant users such as government-sponsored organizations is apparent and unresolved. (But hopefully it will be resolved in the new URAR appraisal ‘form’.)
To sum up: “market conditions” must include any seasonal change for a time adjustment. This is automatic when selecting the Competitive Market Segment (CMS)©
The goal is the signal (the change in price levels), not the “noise” from individual property differences.
Time series analysis is a well-developed body of knowledge in economics and business analytics. It is the best practice, where sufficient sales (say 7-9) are available.
Legacy appraisal “adjustment” practices are inappropriate using time-series analyses. This includes paired-sales and grouped-analysis. Compiled housing price indexes (last resort) do not correlate to any specific market segment. [See the 21 part series on market conditions analysis in this Analogue Blog].
Appraiser judgment is needed for the time span, trend changes, and outlier handling. Visual (graphs) and simple regression software is available as open source, very fast, easy to use, and easy to work with the new URAR and available residential software.