Generally, the decimal type because it allows expression of decimal numbers
with equal precision across all the values it represents and it gives enough
magnitude for all financial operations. Floating point numbers have
approximate precision for many values they are capable of representing -
i.e. 'rounding errors'
Richard
