I don't think there is a usable formula for the exact distribution of your quantity, but $n/m$ times your quantity has approximately a chi squared distribution on $n-1$ degrees of freedom. (After multiplication by $n/m$ your quantity is the Pearson chi squared test statistic, so a lot is known about it.) If $m/n\to\infty$ and also $n\to\infty$, you can probably use a normal approximation to this chi squared distribution.