Hey bob000.
Are the models for the errors distributed according to a Brownian motion/Wiener process?
An asset start with value V.
Each year the asset depreciates by (D plus or minus e)% of its remaining value plus or minus.
The expected value of the asset in year n is V*(1-D)^n.
What is the expected error for year n?
I know that if depreciation is contant plus or minus e' the error accumulates as e'*sqrt(n). I am not sure how to calculate error accumulation if the error is a proportion of the remaining asset. Would it be sqrt[(e*V)^2 + (e*V*(1-D))^2 + (e*V*(1-D)^2)^2 + ...] ?