Anecdotal - I just completed a CA on an SFR that's partially built and which has a 4ac site with a grove in a semi-rural area of San Diego County. It was a complicated appraisal problem overall, the particulars of which I won't go into here. But I had great land sale comparables located in the subject neighborhood, and physical depreciation wasn't a factor.
Without engaging in any hocus-pocus my CA and SCA came in so close to each other that I don't like it.
I was given a 2018 appraisal on the property, performed by an appraiser who put considerable thought into his analysis and there was a lot of overlap in the costs when considering we were using different cost references. One big difference is that I opined to an underlying site value that was 40% lower than the other appraiser. That's because I used actual site sales data and he did "extraction" despite the easy availability of comparable site sales data.
I work in Southern Calif, including some of the urban areas, and a certain percentage of my book of work consists of land appraisal assignments or proposed construction. I *rarely* run into an SFR assignment where there actually is no relevant site sales data. And the reason I say that is because the geographic range for SFR parcels in an urban/suburban setting is usually far wider than the geographic range for the finished or existing home. Except perhaps for the projects with significant common elements you usually don't need site sales in your immediate neighborhood to figure out what the site values are.
Put it this way: If your subject was built out as a subdivision home in the middle of a 400-unit subdivision and it burned down and your assignment was to value a vacant site in the middle of that subdivision you would usually not have to resort to land value by extraction in order to value that lot. You can usually find vacant land sales data to use and then make location adjustments if/when necessary. Which often it won't be. Not even in the urban areas.
