I am running a MySQL 5.7.30-0ubuntu0.16.04.1-log Server where I have the option of saving in char(4) or in smallint(5, unsigned). There will be a primary index on the column and the key will be used as a referrence accross tables.
What is faster? Char or Int?
Unsigned SMALLINT values use two bytes and have values in the range [0, 65535]. CHAR(4) values take four bytes. So, indexing SMALLINT values will make for a smaller index. Smaller is faster. Plus indexes on character columns usually have all sorts of character-set and case-insensitivity monkey business built in to them, which also takes time and space.
But, for a table with at most 65K rows, the effect of this choice will be so small you'll have trouble measuring it. If you build something that's hard to debug, you'll spend your precious time and ten thousand times as much computer time debugging it than it will save.
Design your tables so they match your application. If you're using a four-digit number use SMALLINT.
The next person to work on your code (even if that person is you a year from now) will thank you for a clear implementation.
And keep in mind that MySQL ignores the number in parentheses on INT declarations. SMALLINT(4), SMALLINT(5), and SMALLINT all mean precisely the same thing. MySQL uses the native processor integer datatypes: TINYINT is an 8-bit number, SMALLINT a 16-bit number, INT a 32-bit number, and BIGINT a 64-bit number. Likewise FLOAT is a 32-bit IEEE 754 floating point number and DOUBLE a 64-bit one. The number of digits SMALLINT(4) is a nod to SQL standards compatibility.