Why is the standard-guaranteed range for Int in Haskell exactly [-2^29; 2^29)?

Why is the range so specific? Why not 29-bits wide or 32 bits-wide, why 30?

What is the reasoning? The standard(1) doesn't seem to give it