The result data type is the same type as that of the numeric_value argument.
If the data type of numeric_value is DECIMAL/NUMERIC with a precision less than 38, then the function increases the precision by 1. For example, when a DECIMAL (4,2) value of 99.99 is rounded to 0 places, the function returns a DECIMAL (5,2) value, 100.00. However, if the precision is 38, then the function only reduces the scale value by 1 unless the scale is 0. For example, the function returns a DECIMAL (38, 36) value of 99.999999999 as a DECIMAL (38, 35) value, 100.00.
For information on the default data type formats, see Data Type Formats and Format Phrases.