To answer your question, I will share some general considerations about the geographic variables in the ACS and via IPUMS USA. I am not sure how you manipulated the variables you shared to arrive at your estimates, but hopefully this information will help you identify what the most appropriate identification strategy is for you moving forward.
The place of residence variables (e.g., STATEFIP, COUNTYFIP, MET2013) report geographic information about a person’s residence at the time of the ACS. You should use these to identify the contemporaneous population of your area of interest (in your case, the Orlando- Kissimmee-Sanford FL metropolitan area).
The migration variables (e.g., MIGMET131) will report information about the person’s residence one year prior. MIGMET131 reports the metropolitan area (using 2013 metro delineations) in which the person previously resided if this metro area can be identified. Note that MIGMET131 is only reported for persons who report living in a different residence one year prior. MIGPLAC1 will identify the state or country of residence one year prior, but not the metropolitan area. You may want to consider using MIGRATE1 in conjunction with MIGMET131 to capture people who have moved whose previous residence was not in an identifiable metro area since the MIGMET131 code “00000” applies to both people who have stayed in the same residence and those who were in an unidentifiable metro area 1 year ago.
If you restrict your analytical sample in any way (e.g., you only look at persons residing in Florida), then your estimates using MIGMET131 would miss anyone who lived in the Orlando metro the year prior but moved out of the state of Florida. Put another way, for an analysis of migration behavior you should only use migration variables (i.e. MIGPLAC1 for state and MIGMET131 for metropolitan area) to restrict your analytical sample to a geographic area.
It is also worth noting that the variables MET2013 and MIGMET131 do not exactly correspond to the official delineations. I will briefly summarize here, but I encourage you to consult the MIGMET131 description for more details. For simplicity, I will refer to MET2013 and current residence, but these principles apply to the migration variables that report on prior residence as well. Metro areas are not directly reported in the ACS public use microdata sample (PUMS) files. Instead, IPUMS derives metro areas using the reported public use microdata area (PUMA), the lowest level of geographic detail reported by the Census Bureau in the original PUMS files. IPUMS only identifies a metro area for a PUMA if the majority of the PUMA’s population resides in the identified metro area; for PUMAs that straddle metro borders but with a majority population living in the metro area, some households identified as living in a metro area will not actually live in that metro area (commission error). In some cases, one or more PUMAs may lie entirely within a metropolitan area and represent most but not all of the metro area; in this situation, households that reside in a metro area but are in PUMAs where the majority of the PUMA’s population do not reside in the metro area will NOT be identified as living in the metro area (omission error). We identify metropolitan areas in the data if the sum of match errors (omission + commission errors) is less than 15%. This level of spatial mismatch error is reported in error variables, including MIGMET13ERR (for MIGMET131).
Finally, I will share that I was not able to replicate the specific estimates that you reported. Because you are interested in examining this by occupation, perhaps you have restricted your analytical sample to only the working age population or employed persons. Be sure to weight your analyses (I see that you included PERWT in your list of variables). Also, I encourage you to examine your unweighted sample sizes before making population inferences – OCC2010 will allow you to examine very specific occupation groups, but they may have too few cases to allow for meaningful interpretation by the time you further cut this by migration status for a specific metro area.