Their table, their analyses based off the data they've collected, their methodology. If we're going to use their table we should be using the same or similar reasoning that table is intended to convey.
Note that their total rate of loss doesn't exceed 80% of cost new, the idea being that even a beater can be brought back into productive use.
In terms of valuing the property and skipping the mechanics for a moment, if we articulate the long form of "REL" that question looks something like:
"given the structure's current condition and considering the composition and market trends in the immediate area, how long do we think these improvements will continue to contribute to the value of the whole, above and beyond the contributory value of the underlying land value?"
In the case of the Culver City property my conclusion was that the answer to that question was 0-5 years, which is exactly what I said in my appraisal report. If located in Inglewood or So Central my answer would be very different.
Now, I wasn't valuing that property for a mortgage. That property was one of several properties I'm valuing in an estate. And I made that comment even though I didn't complete a Cost Approach for that assignment. I did complete a land sale analysis (using real land sales) in order to support my HBU analysis. The question occurred to me (early on) during my research so I simply continued to pull on the dangling thread until I found out what I needed to know.
Needless to say, if that appraisal went to a lender and they became aware of the fact that - depending on what type of borrower they were dealing with - they might be making what amounts to a land loan at a 95% LTV it might have been of effect on their loan decision. Or not. Conversely, if I had been appraising that structure in Inglewood and I expressed the opinion that the REL was 25 years or more that wouldn't have been an unreasonable opinion for me to express, either.
The point being that if you use the same methodology M&S uses with their table then it's easy to square your CA with what's actually happening in the market. You only have problems with the CA when you mix-n-match their table with a different equation than it's intended to be used with.