The reason that CO2 increases lag temperature increases is that they are a result, not a cause of global warming. This is demonstrated by the same lag being repeated in multiple deglaciation cycles in the paleoclimate data. This lag and its irresistible logical implications are very inconvenient for the AGW argument.
A discussion of the mechanism is presented here. http://www.rocketscientistsjournal.com/2006/10/co2_acquittal.html. The basic idea is that the relationship between temperature levels and CO2 levels is well explained by the solubility curve of CO2 in water, which is a function of temperature.
The author addresses criticisms of his thesis by Gavin Schmidt here. http://www.rocketscientistsjournal.com/2006/11/gavin_schmidt_on_the_acquittal.html, and is worth reading if you are of a scientific bent of mind.
Regards,
In any case, it would be only fair to subject this theory to the same standard as I'd want for the other, namely practically testable predictions that can be compared to other models.
For the record, I consider figure 10 quite suspect. It claims to show cross-correlation between the two signals for delays up to 400K years. The two signals are not very far from saw tooth waves with the same 100K year period and the cross-correlation of two such pure waves will not behave qualitatively as figure 10 indicates. It will dip down through zero, go pretty far negative then swing back up at the end of the period. Even for noisy signals such as these are, I'd expect it to dip at about 50K years but it hangs pretty steady instead.