Not causation, agreed. It's still correlation though. If so, one could be used as a predictor for the other - even without a hint to the actual cause(s).
If he can find predictive power in it, bully to him, but do a couple of points in year's worth of data sound like a statistically significant correlation? We can correlate all manner of things if we cherry-pick our data to fit.