Statistical inefficiency Markov chain

Question

I am trying to understand Markov chain processes and have trouble understanding the statistical inefficiency related to Markov chains. What i have read is that often in practice states along the Markov chain are separated by some fixed number ns, chosen so that there is effectively no correlation between the states used. The number ns is called statistical inefficiency. How is that a measure of inefficency? What does large values of ns means compared to high values of ns?