Markov's inequality is a probabilistic inequality. It provides an upper bound to the probability that the realization of a random variable exceeds a given threshold.
The proposition below formally states the inequality.
Proposition Let be an integrable random variable defined on a sample space . Let for all (i.e., is a positive random variable). Let (i.e., is a strictly positive real number). Then, the following inequality, called Markov's inequality, holds:
Reading and understanding the proof of Markov's inequality is highly recommended because it is an interesting application of many elementary properties of the expected value.
First note thatwhere is the indicator of the event and is the indicator of the event . As a consequence, we can writeNow, note that is a positive random variable and that the expected value of a positive random variable is positive:Therefore,Now, note that the random variable is smaller than the random variable for any :because, trivially, is always smaller than when the indicator is not zero. Thus, by an elementary property of the expected value, we have thatFurthermore, by using the linearity of the expected value and the fact that the expected value of an indicator is equal to the probability of the event it indicates, we obtainThe above inequalities can be put together:Finally, since is strictly positive we can divide both sides of the right-hand inequality to obtain Markov's inequality:
This property also holds when almost surely (in other words, there exists a zero-probability event such that ).
Suppose that an individual is extracted at random from a population of individuals having an average yearly income of $40,000.
What is the probability that the extracted individual's income is greater than $200,000?
In the absence of more information about the distribution of income, we can use Markov's inequality to calculate an upper bound to this probability:Therefore, the probability of extracting an individual having an income greater than $200,000 is less than .
Markov's inequality has several applications in probability and statistics.
For example, it is used:
to prove Chebyshev's inequality;
in the proof that mean square convergence implies convergence in probability;
to derive upper bounds on tail probabilities (Exercise 2 below).
Below you can find some exercises with explained solutions.
Let be a positive random variable whose expected value is
Find a lower bound to the probability
First of all, we need to use the formula for the probability of a complement:Now, we can use Markov's inequality:Multiplying both sides of the inequality by , we obtainAdding to both sides of the inequality, we obtain
Thus, the lower bound is
Let be a random variable such that the expected value exists and is finite.
Use the latter expected value to derive an upper bound to the tail probabilitywhere is a positive constant.
By Markov's inequality, we have
If you like this page, StatLect has other pages on probabilistic inequalities:
Please cite as:
Taboga, Marco (2021). "Markov's inequality", Lectures on probability theory and mathematical statistics. Kindle Direct Publishing. Online appendix. https://www.statlect.com/fundamentals-of-probability/Markov-inequality.
Most of the learning materials found on this website are now available in a traditional textbook format.