micro: CBSA split by state

[bdecon]

When mapping CPS data to metro areas, one problem is that censored CPS observations are not random. For example, the DC metro area includes a small portion of West Virginia that is too small to disclose in the CPS. So calculating something for the DC metro area using public use CPS files generates a result that is not actually showing the DC metro area, it is showing the DC, MD and VA portion of the metro area.

There is a nonstandard set of geographic areas that the CPS can tell us about. By combining the CBSA variable and the State variable, we can come up with small geographic areas. If one of those areas has too few observations, we can group it with other parts of its metro area. If the area has no observations, we can remove it from the map.