In the Cox-Ingersoll-Ross or CIR Model, Interest Rates Are Specified $d r _ { t } = k \left( \theta - r _ { t } \right) d t + \sigma \sqrt { r _ { t } } d W _ { t }$

In the Cox-Ingersoll-Ross or CIR model, interest rates are specified by the following stochastic process: $d r _ { t } = k \left( \theta - r _ { t } \right) d t + \sigma \sqrt { r _ { t } } d W _ { t }$ One attractive feature of this process relative to the Vasicek interest rate process $d r _ { t } = k \left( \theta - r _ { t } \right) d t + \sigma d W _ { t }$ is that

A) Interest rates are always non-negative in CIR while they may be negative in the Vasicek model.
B) There are parameter restrictions which guarantee non-negative stochastic interest rates in the CIR model, but there are no such restrictions possible in the Vasicek model.
C) It has extra parameters, so can fit observed yield curves better.
D) It allows for imperfect instantaneous correlation between rates of different maturities, whereas in the Vasicek model, they are perfectly correlated.