Choosing the appropriate precision for decimal(x,y)

Answer 1

For 4 decimal places, use money if you can

Reason: it's faster then decimal but is exact

Money, float, decimal article

Edit: based on comment

Interesting argument. I follow both blogs.

Note the examples here: different precision based on datatypes and precedence

declare @d decimal(19,4), @ddiv decimal(19,4)

set @d = 12.39
SET @ddiv = 1000

select (@d/@ddiv)*@ddiv, (@d/1000)*1000
--Gives   12.390000    12.390000000

Explained by my answer here to "T-SQL Decimal Division Accuracy"

Now, this is interesting. Decimal, different answers

declare @d decimal(19,4), @ddiv decimal(19,4)

set @d = 12.39
SET @ddiv = 1001

select (@d/@ddiv)*@ddiv, (@d/1001)*1001
--Gives   12.390000    12.389999622

Back to money: same answer for each

declare @d money, @ddiv money

set @d = 12.39
SET @ddiv = 1001

select (@d/@ddiv)*@ddiv, (@d/1001)*1001
--Gives   12.3123    12.3123

Moral: store in SQL, process in client languages...

Answer 2

The main reason to use a smaller data precision, despite using the same amount of storage space, is to convey meaning to future users of the system. Which is also why it's important to use appropriate data types - e.g. DECIMAL(15,4) for numbers, MONEY for money.