A new Markov model for evaluating the ALR dual-threshold policy