Labour's share of income in the US is at the low end of its usual range of variation:
The long-run stability of this ratio plays an important role in macroeconomic modelling, namely the widespread use of the Cobb-Douglas functional form for production functions.
Over the past 25 years, the ratio of (exports + imports)/GDP has more than doubled. Shouldn't this have had some some sort of effect on the composition of output and factor income?
It's very likely that it has. If the US has a comparative advantage in industries that make use of high-skilled workers, than we'd expect that the share of income that went to unskilled workers would decline, while that of skilled workers would increase. Put the two together, and the effect on the labour share of income may very well be a wash. (Although the effect on income inequality would certainly be noticeable).
The apparent stability of the labour share of income may be masking an important transition in the US economy, one that most macroeconomic models are unable to capture.