According to Merritt Ruhlen, over 50% of occurrences of phonemic labialization are applied to dorsal (velar and uvular) consonants, while coronals are usually left out. There are a few families where this doesn't apply because labialization results from historically rounded vowels, such as the Northwest Caucasian languages, but it appears to exist most of the time. Why is this?
The explanation is based on phonetics. A back (velar or uvular) constriction causes the second formant to be low, as does rounding. Together, labiovelars have a very low F2, and that results in a very distinctive sound. A coronal consonant on the other hand has a high F2, but adding rounding lowers F2, so the resulting sound is perceptually less distinctive. Ohala & Lorenz "the story of w" explains the acoustic basis for labiovelar preferences (though not including [tʷ] compared to [t])
Labialisation has nothing to do with a low F2. If it was the case laryngeal and pharyngealised consonants, which have also a low F2, will know in the same manner a high frequency of labialisation. But such observations have not been noticed, the perception is not at stake here. So we have to explain that in another way.
According to me, the answer is probably a articulatory problem. The sequences [ku] / [kw] or [gu] / [gw] cannot stand together, because they are homorganic (same place of articulation: dorsal). In consequence, a coalescence or a assimilation phenomenon will occur more easily, facilitating the labialisation of these consonants.
Another point, as the rounding posterior vowels and velar consonants are more frequently found among the world languages, so, automatically, labialised velar will be statistically more representative in comparison to others.