RATE function (and FV,NPER,PMT,PV) PayType parameter

Hi,

In our RATE() function we have a PayType parameter that is said to
designate whether payment is due at end of period or start of period,
but we do not explicitly restrict the value to be either 0 or 1. Also
ECMA/MOOXML documents the parameter to be the "timing of the payment,
truncated to integer" and be either 0 or 1, and I guess that is what
applications implemented. However, Excel does not constrain the
parameter to {0;1} but instead yields different results for different
fractional values of PayType, also values >1.0 and <0.0 are possible.

I tend to explicitly allow this and change the "boolean" {0;1} behavior
to an offset (in periods) to the end date of a payment period towards
the start date. However, this must correspond with functions FV(),
NPER(), PMT() and PV().

Opinions on this? Mathematical constraints? Insights on what Excel
really does?

  Eike

-- 
Automatic string conversions considered dangerous. They are the GOTO statements
of spreadsheets.  --Robert Weir on the OpenDocument formula subcommittee's list.