The fact that SYSTABCOL.max_identity is a BIGINT seems to imply negative autoincrement values are possible, but they are not... all that the BIGINT data type does is needlessly cut the autoincrement range in half. Here's a demonstration...

(note that sa_reset_identity accepts negative values but get_identity does not return them)

-----------------------------------------------------------------------------------------
-- SYSTABCOL.max_identity is BIGINT so it can be set to a negative number.

CREATE TABLE t ( pkey BIGINT NOT NULL DEFAULT AUTOINCREMENT );

CALL sa_reset_identity ( 't', new_identity = -5 );

CHECKPOINT;
SELECT max_identity 
  FROM SYSTABCOL INNER JOIN SYSTAB ON SYSTABCOL.table_id = SYSTAB.table_id 
 WHERE SYSTAB.table_name = 't';

max_identity
-5

-----------------------------------------------------------------------------------------
-- A negative SYSTABCOL.max_identity raises an exception and is automatically set to zero.

INSERT t VALUES ( DEFAULT );

Column 'pkey' in table 't' cannot be NULL
SQLCODE=-195, ODBC 3 State="23000"

CHECKPOINT;
SELECT max_identity 
  FROM SYSTABCOL INNER JOIN SYSTAB ON SYSTABCOL.table_id = SYSTAB.table_id 
 WHERE SYSTAB.table_name = 't';

max_identity
0

-----------------------------------------------------------------------------------------
-- Positive values in SYSTABCOL.max_identity work OK.

INSERT t VALUES ( DEFAULT )
COMMIT;
SELECT * FROM t;

pkey
1

-----------------------------------------------------------------------------------------
-- GET_IDENTITY returns UNSIGNED BIGINT.

SELECT EXPRTYPE ( 'SELECT GET_IDENTITY ( ''t'' )', 1 );

EXPRTYPE('SELECT GET_IDENTITY ( ''t'' )',1)
unsigned bigint