COALESCE in SQL Server vs COALESCE in MySQL

As we all know COALESCE is an ANSI SQL standard function. It has the same functionality across different RDBMS (ie) It returns the first NOT NULL value from list of values.

Consider the following data setup

CREATE TABLE TableA (customerID varchar(10), salary int);
INSERT INTO TableA (customerID, salary)
VALUES
    ('A1', 100),
    ('A2', 200),
    ('A3', 300),
    ('A4',400);

CREATE TABLE TableB (customerID varchar(10), rate int);
INSERT INTO TableB (customerID, rate)
VALUES
    ('A1', 2),
    ('A2', 3),
    ('A3', 4);

Query :

SELECT t1.customerID,
       COALESCE(t1.salary * t2.rate, 'NA') AS salary
FROM TableA t1
LEFT JOIN TableB t2
    ON t1.customerID = t2.customerID

In SQL Server when I ran the above code, it generates the following error

Msg 245, Level 16, State 1, Line 64 Conversion failed when converting the varchar value 'NA' to data type int.

It's because COALESCE function will convert the NA to Integer since Integer has higher precedence than varchar.

When we run the same code in MySQL it works. Here is the Demo

My question is, Why an ANSI SQL function implemented in two different ways in two different RDBMS(SQL Server & MySQL). Till now my thought was ANSI SQL function behaves in same way in all RDBMS!

Solution

IMO, The COALESCE is implemented in the same way in both databases as achieving this goal:

Returning first Non-Null parameter

But there is a policy in SQL Server that is not in MySQL, something like:

Just same values can union as one column

Above policy is applied in many queries:
e.g.:

SELECT 1 AS column1
UNION ALL
SELECT 'c'

SELECT 
     CASE A 
         WHEN 1 THEN 1
         WHEN 2 THEN 'c'
     END
 FROM ...

....

I think this is not related to implementation of COALESCE it's related to query policies of databases.