Character-string and item size
For items described with a PICTURE clause, the size of an elementary item is expressed in source code by the number of character positions described in the PICTURE character-string and a SIGN clause (if applicable). Storage size, however, is determined by the actual number of bytes the item occupies as determined by the combination of its PICTURE character-string, SIGN IS SEPARATE clause (if specified), and USAGE clause.
For items described with USAGE DISPLAY (categories alphabetic, alphanumeric, alphanumeric-edited, numeric-edited, numeric, and external floating-point), 1 byte of storage is reserved for each character position described by the item's PICTURE character-string and SIGN IS SEPARATE clause (if applicable).
For items described with USAGE DISPLAY-1 (category DBCS), 2 bytes of storage are reserved for each character position described by the item's PICTURE character-string.
For items described with USAGE NATIONAL (categories national, national-edited, numeric-edited, numeric, and external floating-point), 2 bytes of storage are reserved for each character position described by the item's PICTURE character-string and SIGN IS SEPARATE clause (if specified).
For items described with USAGE UTF-8 (category UTF-8), when the BYTE-LENGTH phrase of the PICTURE clause and the DYNAMIC LENGTH clause are not specified in the item’s definition, 4 bytes of storage are reserved for each character position described by the item's PICTURE character-string. Note, however, that due to the varying-length nature of UTF-8 characters, many UTF-8 strings of the specified character length can be represented using fewer bytes than the maximum reserved. In such a case, any unused bytes in the data item are padded with UTF-8 blanks.
For internal floating-point items, the size of the item in storage is determined by its USAGE clause. USAGE COMPUTATIONAL-1 reserves 4 bytes of storage for the item; USAGE COMPUTATIONAL-2 reserves 8 bytes of storage.
Normally, when an arithmetic item is moved from a longer field into a shorter one, the compiler truncates the data to the number of digits represented in the shorter item's PICTURE character-string by truncating leading digits. For example, if a sending field with PICTURE S99999 that contains the value +12345 is moved to a BINARY receiving field with PICTURE S99, the data is truncated to +45. For additional information, see USAGE clause.
The TRUNC compiler option can affect the value of a binary numeric item. For information about TRUNC, see TRUNC in the Enterprise COBOL Programming Guide.