Amazon pays little tax because of all the losses it posts. One potential explanation for this might be that small cap companies have higher profits than large caps. If that was true they would pay more tax collectively. I have heard before that small caps tend to grow more but are higher-risk, and since there is a selection bias against companies that collapsed last year, I imagine you would find that of the companies paying tax this season small caps would be showing more growth. So, the causal chain I'm proposing would be:

Small caps have higher variance than large caps + Companies that go out of business aren't included in statistics -> Year-end statistics show small caps doing better than large caps.

Then,

Companies that do better on paper pay more tax + Small-caps do better on paper at year end -> small caps pay more tax than large caps.