Mean dependence

In probability theory, a random variable ${\displaystyle Y}$ is said to be mean independent of random variable ${\displaystyle X}$ if and only if its conditional mean ${\displaystyle E(Y\mid X=x)}$ equals its (unconditional) mean ${\displaystyle E(Y)}$ for all ${\displaystyle x}$ such that the probability density/mass of ${\displaystyle X}$ at ${\displaystyle x}$, ${\displaystyle f_{X}(x)}$, is not zero. Otherwise, ${\displaystyle Y}$ is said to be mean dependent on ${\displaystyle X}$.

Stochastic independence implies mean independence, but the converse is not true.^[1][2]; moreover, mean independence implies uncorrelatedness while the converse is not true. Unlike stochastic independence and uncorrelatedness, mean independence is not symmetric: it is possible for ${\displaystyle Y}$ to be mean-independent of ${\displaystyle X}$ even though ${\displaystyle X}$ is mean-dependent on ${\displaystyle Y}$.

The concept of mean independence is often used in econometrics^{[citation needed]} to have a middle ground between the strong assumption of independent random variables (${\displaystyle X_{1}\perp X_{2}}$) and the weak assumption of uncorrelated random variables ${\displaystyle (\operatorname {Cov} (X_{1},X_{2})=0).}$

Further reading

• Cameron, A. Colin; Trivedi, Pravin K. (2009). Microeconometrics: Methods and Applications (8th ed.). New York: Cambridge University Press. ISBN 9780521848053.
• Wooldridge, Jeffrey M. (2010). Econometric Analysis of Cross Section and Panel Data (2nd ed.). London: The MIT Press. ISBN 9780262232586.

References

1. Cameron & Trivedi (2009, p. 23)
2. Wooldridge (2010, pp. 54, 907)