Not in reality. Rapidly fluctuating construction costs and estimating depreciation render it totally unreliable.
Regardless short term fluctuations in prices of materials, I've never seen a building contractor who does not actually charge the going rate locally without regard to the cost except to make sure there is a margin of profit. So, despite lumber going from $1100/k to $450/k, I've yet seen a single bid that I thought was lower than it was last year or the year before. In fact, here the per SF cost is rising and has continued to rise for 3-4 years. The builder who was building for $120/SF in 2018 is now $150-180/SF and up. That's one reason we are seen more new Barndominiums being built. They are running less than $100/SF. My neighbor told me his cost less than $100/SF and he finished it about 2 months ago.
Secondly, the cost book is an excellent way to justify the cost related adjustment for unique items. It's a way to estimate the RCN of a shop or pool or other item in the analysis. The historical cost index also gives some idea of the inflation of an item or home when doing a retrospective appraisals.
The CA is not created in a vacuum. And if the RCN (not the DCN) plus land exceeds the SA then it acts as an upper level where it makes sense to the buyer to build, not buy existing housing. That does not mean the CA should always be higher than the SA. But all three approaches are based upon market data, and they are interlocking.
When I started, one mentor said he was not comfortable using only a single approach. If nothing else, if there was a huge difference in the derived value, it was useful to go back and ID why and see if you have not made a math error.
funny is that reporting a CA
Only FNMA et al limits themselves to the SA. And they have so many other parameters - hoops to jump through so to speak - that they felt the risk is worth taking. OTOH, they also demand you support each and every adjustment and again, for those oddball items - the she shed, the shop building, the horse barn, "extraction" from the market is pretty much a fairy tale of pretending to extract from the market data and just pulling a number out of your backside.