Request for a WPIC Datatype in Enterprise PL/I for z/OS

We started to adapt our PL/I programs to support Unicode data. For this purpose we use the PL/I Datatype WCHAR (WIDECHAR) instead of CHAR for input/output and processing of all kinds of text-data.
This means that all such text data is stored in UTF-16 in the program as well as in
DATASETS and in DB/2 Databases. (In DB/2 this data is stored in
GRAPHIC or VARGRAPHIC columns.)
One of the biggest problems in this endeavor is the lack of a Unicode
PICTURE type "WPIC" to store formatted numerical data in UTF-16
In many situations we need to create records containing numeric data in
text (zoned) format (for printing or data exchange or storing data as
text string in a database column ...).
For this we use structures with CHAR and PIC like

DCL 1 INFO,
3 ITEMNAME CHAR(20),
3 ITEMCOUNT PIC'(3)Z9',
3 AMOUNT PIC '(9)Z9.V99';

WRITE FILE(XY) from(INFO);

If we now store text strings in Unicode and if we use
DCL 1 INFO,
3 ITEMNAME WCHAR(20),
3 ITEMCOUNT PIC'(3)Z9',
3 AMOUNT PIC '(3)Z9.V99';
WRITE FILE(XY) from(INFO);

then this results in a mixture of UTF-16 and EBCDIC which is not useable for our purposes.

If PL/I would support a WPIC data type, which stores the numbers and charachters in UTF-16, then we could use

DCL 1 INFO,
3 ITEMNAME WCHAR(20),
3 ITEMCOUNT WPIC'(3)Z9', /* fictional WPIC DATATYPE
invented by AMOS AT */
3 AMOUNT WPIC '(3)Z9.V99';
WRITE FILE(XY) from(INFO);

and this would result in a consistent UTF-16 String.

So the variable AMOUNT with type WPIC '(3)Z9.V99' would use 14 Bytes
(to represent 7 characters/numbers).
AMOUNT=12.34;
would have the hex representation
0020 0020 0031 0032 002E 0033 0034

The WPIC datatype would also allow converting from packed to zoned format by structure assignment, which we also use quite frequently

DCL 1 packedINFO,
3 ITEMNAME WCHAR(20),
3 ITEMCOUNT dec fixed(5),
3 AMOUNT dec fixed(11,2);

DCL 1 INFO,
3 ITEMNAME WCHAR(20),
3 ITEMCOUNT WPIC'(3)Z9',
3 AMOUNT WPIC '(3)Z9.V99';

info=packedinfo, by name;

Could you please examine, if it is possible to implement such a datatype in Enterprise PL/I for z/OS? This datatype would drastically reduce the effort for adapting our programs for unicode and allow us to keep our code concise.