The suggestions in this answer assume that a new table must be created and that the source tables are declared with columns in the same sequence.
(1) You can insert one table at a time, without the need for UNION ALL:
-- código #1
SELECT coluna_1, coluna_2, ..., coluna_n
into DATATRAN_CSV
from DATATRAN2013_CSV;
INSERT into DATATRAN_CSV (coluna_1, coluna_2, ..., coluna_n)
SELECT coluna_1, coluna_2, ..., coluna_n
from DATATRAN2014_CSV;
INSERT into DATATRAN_CSV (coluna_1, coluna_2, ..., coluna_n)
SELECT coluna_1, coluna_2, ..., coluna_n
from DATATRAN2015_CSV;
INSERT into DATATRAN_CSV (coluna_1, coluna_2, ..., coluna_n)
SELECT coluna_1, coluna_2, ..., coluna_n
from DATATRAN2016_CSV;
In this way, fewer memory resources are used.
(2) If you want to merge the source tables, before creating the new table, here's another option:
-- código #2
with cteDATATRAN as (
SELECT * from DATATRAN2013_CSV
union all
SELECT * from DATATRAN2014_CSV
union all
SELECT * from DATATRAN2015_CSV
union all
SELECT * from DATATRAN2016_CSV
)
SELECT *
into DATATRAN_CSV
from cteDATATRAN;
Note that you must use UNION ALL to get the complete union. If you use only UNION, and the process slows down, repeated lines are discarded.
(3) To have absolute control, I suggest you do not use * in the column list of the source tables, but rather define the column names:
-- código #3
with cteDATATRAN as (
SELECT coluna_1, coluna_2, ..., coluna_n
from DATATRAN2013_CSV
union all
SELECT coluna_1, coluna_2, ..., coluna_n
from DATATRAN2014_CSV
union all
SELECT coluna_1, coluna_2, ..., coluna_n
from DATATRAN2015_CSV
union all
SELECT coluna_1, coluna_2, ..., coluna_n
from DATATRAN2016_CSV
)
SELECT *
into DATATRAN_CSV
from cteDATATRAN;