In the absence of good comps in a neighborhood, you should first ask, why?
* Are there (or have there been ) any recent listings in the neighborhood, or are properties sitting on the market for a long time and not selling (or cancelled/expired listings)?
This should give you a clue as to the desirability (demand) for the neighborhood or area.
....you may find there are ample sales in adjacent neighborhoods, but lingering active listings (or cancelled/expired listings), in the Subject neighborhood. This may point to the Subject being in a less desirable neighborhood, or may simply be due to owners holding on to their houses in the neighborhood and little are offered for sale.
or
Are the recent sales in the neighborhood , only 1,400sf ?
...and if so, is that because an 1,800sf house is overbuilt for the neighborhood?
or .....is it just that 1,800sf houses are more desirable and less offered for sale?
....maybe an 1,800sf house is highly desirable in the area, but their have not been any available for sale recently.
In reality, an 1,800sf house does not sound overbuilt for most neighborhoods, but this can be the case in larger gla's (example. 3,000sf house, in a neighborhood with mainly 1,400sf houses).
Over the years, I have had properties offered to me that were significantly larger than what is typical for an area, with the Seller/s trying to push how much better those properties were, due to the superior gla. But, for me, that just contributed to more risk, since there are typically less buyers for the "atypical" house in a neighborhood. I typically want to have the typical house (in primary characteristics) in a neighborhood, for a relatively low price and then make it superior in condition and quality.
And, contrary to some opinions, an appraiser, or should I say, a Good appraiser, will not look at it on a $/sf basis, especially not prior to adjusting for other factors. Example - A run-down, unmaintained 1,000sf house, with no pool and located on a high traffic street, should not be worth the same as a fully remodeled 1,000sf house with twice the lot size and a pool, etc.
I suppose you can look at $/sf on a more general basis for an area (assuming you have many datapoints), but then you still want to take into account the Subject size vs the typical size for the area. Because, if you are overbuilt for the neighborhood, then the surplus gla will have less value/sf than the typical gla. So, if you are an 1,800sf house in an area of 1,400sf houses, the 400sf additional sf, may be worth less than the first 1,400sf. And you would be overvaluing the 1,800sf by just doing a basic $/sf calculation, without taking that into account. ..... Makes sense?
Ok, this might be more info than you wanted ...but it is important.
Here are some other techniques to follow to help evaluate a property with little local data.
You should look at it from multiple sides:
* Look at historical sales (go back years if you need to) and see if properties in the neighborhood typically sell fast, and note their gla's (1,400 or 1,800sf, etc), and get a feel for the Subject's value at that time. Then you can "trend" the appreciation of the market from that time, to estimate what the appreciation rate may have been from then till now. So, if you think 3 years ago the value should $100k and you estimate 10% appreciation, you may decide the current value may be around $110k.
* Also, look at adjacent neighborhoods. Find recent sales of similar properties in similar neighborhoods. That should give you a fair indication of value for the Subject. If you can't find any, then use relatively similar neighborhoods to find sales and adjust for their sizes, year built, etc.
Basically, you should be taking multiple factors into account in the areas you are finding recent sales
* Is that neighborhood similar to the Subject, similar demand and appeal?
* Are homes of similar characteristics to the Subject neighborhood (i.e. size, year built, bed/bath count, quality, condition, lot size, etc).
In general, you are looking for recent data of similar properties, and if you can't find similar properties, you find the best data you can and try and take any differences into account (location, size, etc). For example, if you only have recent sales in a superior neighborhood, where homes sell for 10% higher than the Subject's neighborhood, then you estimate the value of your house and adjust it down by 10% for being in an inferior neighborhood. and you can do the same for any characteristic (superior or inferior). That's actually how a good appraiser should be doing it. And yes, you may use $/sf, but just keep in mind the potential pitfalls of that and price that in to your estimate.
Also, to be clear, I am a 25 year certified Appraiser, Investor and Broker