Yule's "Nonsense Correlation" Solved!

TL;DR

This paper mathematically confirms Yule's 1926 empirical finding of 'nonsense correlation' by analytically calculating the second moment of the empirical correlation coefficient between two independent Wiener processes, revealing its high volatility.

Contribution

The paper provides the first explicit analytical calculation of the second moment of the correlation coefficient for independent Wiener processes, confirming the nature of 'nonsense correlation' and offering formulas for higher moments.

Findings

Standard deviation of correlation coefficient is nearly 0.5.

Correlation is heavily dispersed and often large in absolute value.

Correlation arises from self-correlation within Wiener processes.

Abstract

In this paper, we resolve a longstanding open statistical problem. The problem is to mathematically confirm Yule's 1926 empirical finding of "nonsense correlation" (\cite{Yule}). We do so by analytically determining the second moment of the empirical correlation coefficient \beqn \theta := \frac{\int_0^1W_1(t)W_2(t) dt - \int_0^1W_1(t) dt \int_0^1 W_2(t) dt}{\sqrt{\int_0^1 W^2_1(t) dt - \parens{\int_0^1W_1(t) dt}^2} \sqrt{\int_0^1 W^2_2(t) dt - \parens{\int_0^1W_2(t) dt}^2}}, \eeqn of two {\em independent} Wiener processes, $W_1,W_2$. Using tools from Fred- holm integral equation theory, we successfully calculate the second moment of $\theta$ to obtain a value for the standard deviation of $\theta$ of nearly .5. The "nonsense" correlation, which we call "volatile" correlation, is volatile in the sense that its distribution is heavily dispersed and is frequently large in absolute value. It is induced because each Wiener process is "self-correlated" in time. This is because a Wiener process is an integral of pure noise and thus its values at different time points are correlated. In addition to providing an explicit formula for the second moment of $\theta$, we offer implicit formulas for higher moments of $\theta$.