The decimal implementation of casting values into primitive integers requires the exponent to be exactly zero. However, it could be more permissive (and useful) and require that the exponent is simply greater than zero, and do some bookkeeping at the end to multiply the result by the appropriate power.