This work is based on the framework proposed by Conrad (1997) to determine the optimal timing of an investment or policy to slow global warming. While Conrad formulated the problem as a stopping rule option pricing model, we treat the policy decision by considering the total damage function that enables us to make some interesting extensions to the original formulation. We show that Conrad's framework is equivalent to minmization of the expected value of the damage function under the stochastic optimal stopping rule. We extend Conrad's model by allowing for policy cost to grow with time. In addition to closed form solution, we also perform Monte Carlo simulations to find the distribution for the total damage and show that at higher quantiles the damage may become too large and so is the risk on the global economy. We also show that the decision to take action largely depends on the cost of the action. For example, in the case of model parameters calibrated as in Conrad (1997) with a constant cost, there is a rather long wait before the action is expected to be taken, but if the cost increases with the same rate as the global economy growth, then action has to be taken immediately to minimize the damage.