A file copybook may show PIC 9(9), but that does not tell you how the value is stored. A COBOL USAGE clause answers that question. It tells Enterprise COBOL whether the field is stored as display digits, binary, packed decimal, native binary, or floating point.
What is the COBOL USAGE clause?
The USAGE clause specifies the format used to represent a data item in storage. If you do not code it for ordinary numeric or alphanumeric fields, COBOL normally treats the item as DISPLAY. That is easy to read, but it is not always the best choice for arithmetic.
Use USAGE deliberately for fields that appear in calculations, counters, subscripts, DB2 host variables, file records, and interface copybooks. A wrong usage can give bad values, wasted storage, slow arithmetic, or a mismatch with another program.
Quick comparison
| Usage | Meaning | Common use |
|---|---|---|
DISPLAY |
External decimal or character representation. | Report fields, text files, SYSIN values, displayable numeric data. |
COMP or BINARY |
Binary storage controlled by the picture size and compiler truncation rules. | Counters, indexes, lengths, record counts. |
COMP-3 or PACKED-DECIMAL |
Packed decimal storage, two decimal digits per byte except the sign nibble. | Money, balances, quantities, packed fields in files and DB2 host variables. |
COMP-5 |
Native binary storage using the full capacity of 2, 4, or 8 bytes. | System interfaces, binary lengths, fields shared with C or assembler. |
COMP-1 and COMP-2 |
Internal floating-point data. These items do not use a PICTURE clause. |
Scientific or engineering values with a wide range of scale. |
DISPLAY usage
DISPLAY stores data in a readable external form. A numeric display field such as PIC 9(5) stores five digit characters. This is useful for printable reports, flat-file fields exchanged as text, and values that must be easy to inspect in a dump or editor.
01 WS-CUSTOMER-NO PIC 9(5) USAGE DISPLAY.
01 WS-CUSTOMER-NAME PIC X(30).The tradeoff is arithmetic cost. When a display numeric field is used in arithmetic, COBOL must work with its numeric value rather than just its printable characters. For heavy calculations, binary or packed decimal fields are usually a better fit.
COMP and BINARY usage
COMP, COMPUTATIONAL, and BINARY are treated as binary forms in Enterprise COBOL. IBM documents the normal storage size as 2 bytes for 1 through 4 digits, 4 bytes for 5 through 9 digits, and 8 bytes for 10 through 18 digits.
01 WS-READ-COUNT PIC S9(9) COMP VALUE 0.
01 WS-TABLE-SUB PIC S9(4) COMP VALUE 1.Binary fields work well for counters, subscripts, loop control, record counts, and lengths. Be careful with fields that leave your program in a file or message. Another program must use the same storage expectation, or the bytes will not mean the same value.
COMP-3 packed decimal usage
COMP-3 is the short form of PACKED-DECIMAL. Packed decimal stores two decimal digits in each byte, except the final half-byte also carries the sign. It is a common mainframe choice for money and business amounts because decimal precision matters.
01 WS-INVOICE-AMOUNT PIC S9(7)V99 COMP-3.
01 WS-TAX-AMOUNT PIC S9(5)V99 COMP-3.
01 WS-TOTAL-AMOUNT PIC S9(8)V99 COMP-3.IBM notes that packed decimal fields use storage most efficiently when they have an odd number of digits. For example, S9(7)V99 has nine digits, so the packed decimal bytes are used neatly with the sign nibble.
COMP-5 native binary usage
COMP-5 is native binary. IBM documents that a COMP-5 item can contain values up to the capacity of the binary field size, not only the value implied by the number of 9s in the PICTURE. This matters for interface fields.
01 WS-BUFFER-LENGTH PIC 9(9) COMP-5.
01 WS-RETURN-CODE PIC S9(9) COMP-5.Use COMP-5 when a called routine, system service, C program, or assembler module expects a native binary value. Do not swap COMP and COMP-5 casually in a shared copybook. Check the compile options and interface rule first.
COMP-1 and COMP-2 usage
COMP-1 is single-precision internal floating point and COMP-2 is double-precision internal floating point. IBM documents COMP-1 as 4 bytes and COMP-2 as 8 bytes. These fields do not have a PICTURE clause.
01 WS-RATE-SHORT USAGE COMP-1.
01 WS-RATE-LONG USAGE COMP-2.Floating point is not the usual choice for currency. Use packed decimal for account balances, invoice totals, interest posted to an account, and other business amounts where exact decimal behavior is expected.
How to choose the right usage
| If the field is... | Usually choose... | Reason |
|---|---|---|
| Printed or read as text | DISPLAY |
Readable and simple for reports, SYSIN, and display files. |
| A business amount | COMP-3 |
Keeps decimal arithmetic suitable for money and packed file fields. |
| A counter or subscript | COMP or COMP-5 |
Binary arithmetic is a good fit for whole-number loop values. |
| A field passed to C or assembler | COMP-5 |
Native binary values match many low-level interface contracts. |
| A scientific measurement | COMP-1 or COMP-2 |
Floating point supports a wide range of scale. |
USAGE at group level
A USAGE clause can be specified at group level, and then it applies to elementary items inside the group. This can be useful, but it also hides detail. Many teams prefer coding usage on elementary numeric fields so copybook reviewers can see the storage format quickly.
01 WS-TOTALS.
05 WS-ITEM-COUNT PIC S9(9) COMP.
05 WS-GROSS-AMOUNT PIC S9(9)V99 COMP-3.
05 WS-NET-AMOUNT PIC S9(9)V99 COMP-3.Do not mix usage casually in file copybooks. If a field is written to a dataset, another job may read the same bytes tomorrow morning. The data definition is part of the file contract.
Common mistakes
Using DISPLAY for every number
DISPLAY is easy to read, but it is a poor default for every calculated number. A record loop that adds totals millions of times should not use display numeric work fields just because the input file is display format.
Using floating point for money
Use COMP-3 for money unless your site has a specific rule that says otherwise. Floating point can represent a wide scale, but finance programs usually need exact decimal behavior.
Changing COMP to COMP-5 without checking interfaces
COMP-5 can use the full native binary capacity of the field. That is useful for interfaces, but it can change behavior compared with picture-limited binary fields. Test boundary values before changing a shared field.
Forgetting TRUNC and ARITH options
TRUNC affects binary items such as COMP and COMP-4. ARITH affects the maximum digits allowed in decimal arithmetic. Review compile options when a numeric field behaves differently after migration.
Related Mainframe Forum guides
For nearby topics, read COBOL Data Division complete guide, COBOL COMP-3 packed decimal, COBOL ARITH compiler option, COBOL TRUNC compiler option, and Working Storage vs Local Storage.
External references
IBM documents the USAGE clause, computational items, and numeric data storage.
FAQ
What is USAGE in COBOL?
USAGE tells COBOL how a data item is represented in storage, such as display, binary, packed decimal, native binary, or floating point.
Is COMP-3 the same as packed decimal?
Yes. In Enterprise COBOL, COMP-3 and PACKED-DECIMAL describe packed decimal storage.
When should I use COMP-5?
Use COMP-5 when a field must hold a native binary value, especially for system interfaces, C calls, assembler calls, lengths, and return codes.
Should money use COMP-1 or COMP-2?
No, not for normal business amounts. Money is usually better defined as COMP-3 because packed decimal fits exact decimal business arithmetic.
