dropping out series with missing observations

[tclark]

I have a problem that is going to involve recursively looping over time to form estimates from a set of series with data available. Over time, the number of series with data will change. So I want to read in all of the series and then, for each time loop, pull the subset of variables with complete data into a block of series that will be used for the computations. I can do this brute force by looping over all series and then putting just those with data available in a new vec[ser]. Is there a more efficient way to do this? Thanks.