1

Consider a random variable $V$ with variance $\sigma_V^2$.

Since the covariance between a random variance and a constant is zero, I think, if $\sigma_V^2=0$, the covariance between $V$ and another random variable, say $U$, should be zero.

That is, $\sigma_V^2=0 \implies \sigma_{VU}=0$ for any random variable $U$.

I am strongly sure. But, I want to check this one more again.

MinChul Park
  • 381
  • If the variance is zero, the variable is constant, so check https://stats.stackexchange.com/questions/267152/correlation-with-a-constant/267155#267155 – Tim May 22 '23 at 06:25

1 Answers1

2

Yes, this is the consequence of the Cauchy-Schwarz inequality: \begin{align} \operatorname{Cov}(U, V)^2 \leq \operatorname{Var}(U)\operatorname{Var}(V). \end{align}

Zhanxiong
  • 18,524
  • 1
  • 40
  • 73
  • 1
    And another way to get the result is from the definition of covariance as the expected value of the product of the deviation of U from its mean with the deviation of V from its mean. Variance of V = 0 means that the deviation of V from its mean is always zero. Thus Cov is the expectation of a product of two functions, one of which is zero. Thus the expectation is zero. – Graham Bornholt May 22 '23 at 03:30