Why are there fewer MSAs identified by the metarea variable in 1990 than there are in later ACS data samples?

I am analyzing the construction sector. When I keep only the observations of people who have occ2010 codes 6210 through 6765 and then count the number of MSAs by using the variable metarea, I only count 250 MSAs in 1990, but there are 284 MSAs in the years 2005-2011. Is there a reason why there are fewer MSAs in 1990 than in the ACS data in the later years? Also, why are there only 62 MSAs in 2003 when there are quite a bit more in other years?

Geographic identification in modern census samples is difficult because the lowest level geographic identification is the PUMA (public use microdata area). PUMA boundaries sometimes straddle official metro area boundaries and thus make it difficult to completely identify each household’s geographic location via metro area. As a result the populations of many metro areas are only partially identified by METAREA codes, and some are completely omitted in some samples. The table on this page lists the percent of each metro area’s population that resides in areas not identified in each sample.

I hope this is helpful. Let us know if you have any additional questions.

Thank you! I sincerely appreciate your help.