Re: [lmi] Question about "Numeric summary" logic

[Greg Chicares]

On 2017-07-30 00:21, Vadim Zeitlin wrote:
> On Sat, 29 Jul 2017 22:44:46 +0000 Greg Chicares <address@hidden> wrote:
[...]
> GC> Thus, for general- and separate- account policies respectively we have
> GC> these two sub-bases:
> GC>   mce_gen_basis: {"Current", "Guaranteed", "Midpoint"}
> GC>   mce_sep_basis: {"Hypothetical", "Zero", "Half of hypothetical"}
> GC> each of cardinality three, which combine mystically to form the seven
> GC> combinations in 'mce_run_basis', of which five are actually used. Surely
> GC> this {2,3,5,7} pattern resulting from interference between federal and
> GC> state regulation must appeal to the Pythagorean in you.
> 
>  This is very impressive but, even more surprisingly, quite understandable,
> thank you! I hope I don't spoil everything by asking a potentially very
> stupid question, but is what you called "Hypothetical" above the same thing
> as is called "full" in the actual code or did I miss another part of the
> pattern?

The inline documentation for set_run_basis_from_cloven_bases()
(below) is supposed to answer such questions, and I think it
does answer this one in the affirmative, though I had to go
back and read it to be sure. IIRC, I wanted a much shorter
and less pompous word than "hypothetical", so I extended
  zero ... half ... ____
and the natural choice seemed to be "full".

"CF" and "GF" are deliberately shared between the disjoint
"general" and "separate" universes. That's so that we can
always speak of a "current" and a "guaranteed" column, those
two being the most important. To make those two align, I had
to torture the underlying notions: the C-G-M basis deals with
interest as well as mortality- and business-expense assumptions,
while the F-Z-H basis signifies only an interest assumption that
more or less replaces the interest assumption inherent in the
C-G-M basis.

/// Illustrations are run on two primary bases:
///
/// - a "general" basis that governs expense charges and general-
///   account rates:
///     {current, guaranteed, midpoint} (here, {C, G, M} for short)
///   where "midpoint" is used only when the illustration reg governs.
///
/// - a "separate" basis that governs separate-account rates:
///     {full, zero, half}  (here, {F, Z, H} for short)
///   where "full" is the undiminished hypothetical rate, and "half"
///   is used only on NASD illustrations that show three rates instead
///   of two.
///
/// To avoid nested loops, it is convenient to represent combinations
/// of those primary bases as a distinct type: a "run" basis for
/// iteration.
///
/// Only these combinations ever arise:
///   {CF, GF, MF, CZ, GZ, CH, GH} actually-used bases
/// of which only these subsets are used:
///   {CF, GF, MF                } illustration reg
///   {CF, GF,     CZ, GZ        } normal NASD
///   {CF, GF,     CZ, GZ, CH, GH} three-rate NASD

BTW, "three-rate NASD" illustrations are no longer used in practice.