sum of grouped PUMA estimates within a state not equalling state estimate

Hi,

I'm working with all the 3-year ACS PUMS files for the state of Texas. I have state-level estimates, and 10 sub-state/regional estimates that are groups of PUMAs. I can't get the sum of the the 10 regional estimates to equal the state-level estimate for all 3-year ACS time periods. The numbers are relatively close: for example for '05-'07 at the state-level I get 5,564589 workers in the TX labor force who meet certain criteria, and 5,347,171 for the sum of the estimates for the10 groups of PUMAs.

Am I doing something wrong, or is there some reason as to why the sums might not add up exactly (perhaps due to the way ACS samples, uses some kind of controls, etc.)?

Has anyone else ran into this before?

Thanks,
Diana