Regressing smoothed probability

[aa88]

Hi,

I am testing whether or not uncovered interest rate parity holds using the Markov Switching model and assuming that there 2 regimes: the "UIP" regime and the "non-UIP" regime. After estimating the model, I used Proc>Make Regime Probabilities Group>Smoothed to get the probability of the UIP regime, let's call it "Prob". The smoothed probability is ranging from 0 to 1, but only asymptotically, so it' can't be considered as a binary time-series variable. Now, if I want to regress this smoothed probability against some regressors, what kind of estimation method should I use? Is it okay to use simple OLS given that "Prob" is not a binary variable?
Thanks in advance.