Actuarial/Mortality Table and cross testing.

I have always used the Up-84 mortality table when doing cross testing on a New Comp Plan.  My plan is failing the average benefits test slightly (69.89%)  I noticed in the administrative system that I use, there is a table called 2017 417 (e) (3).    If I use that table the Average Benefits test will pass at 70.20%.    The plan document does not state which table to use, so, Is it ok to use this table when running the test?   

Yes, for a 2017 test.  Some may argue that it could be used for any year.  I haven't looked at that issue in a long time so maybe somebody else has the time to look that up.

the boring Math details

consider the following. the first uses UP 1984, fails

the second uses 1983 IAF passes (NHCE e bar is greater than the HCE E BAR)I always recommend using 1983 IAF if cross testing DC plan only. certainly one of the tables that is 'acceptable' if that makes you feel more comfortable.

notes: accumulation factor = 1.085 ^ 5 or 1.085^ 21 depending which person

equiv amt = total allocation * accum factor

annual benefit = equiv amt / APR *12

E Bar = annual benefit / comp

these are the extreme ends of the mortality tables - APR = 95.38 vs 115.39

it helps to impute disparity

might make a difference whether you used age definition nearest or age definition last

If you do not impute permitted disparity changing tables will not do anything for you..

I don't think that's true, because if you have employees past NRA, the actual table affects how much the succeeding APRs decrease after that age.  It's not like a 66 year old's APR would be always, say, 97% of the age 65 APR.  Which, if we're splitting hairs on EBARs out to five decimal places (thousandths of a percent), can make a difference.

Yes, the table won't affect the results unless imputing PD or with participants past NRA (otherwise you are just dividing the projected contribution by a constant [test age APR] for all participants).

IME depending on demographics it helps to use either the shortest or the longest table.  In practice I've used 71 GAM, 83 IAF,  or the applicable table.  I can't recall where anything in between ever helped, but it's been a long time since I experimented with this.