Common Priors and Markov Chains

The type function of an agent, in a type space, associates with each state a probability distribution on the type space. Thus, a type function can be considered as a Markov chain on the state space. A common prior for the space turns out to be a probability distribution which is invariant under the type functions of all agents. Using the Markovian structure of type spaces we show that a necessary and sufficient condition for the existence of a common prior is that for each random variable it is common knowledge that all its joint averagings converge to the same value.