I just reread your question and missed the "over 5 years of data" part.
The way households are sampled, a HU could be sampled again 5 years later (but not 6 or 7 because of rotating sub-frames). But there is no way of identifying if two observations 5 years apart came from the same HU.