Using sipp data, I am examining whether those at the lower end of income distribution are more likely to move than their higher income counterparts. To fully utilize sipp data that contains monthly information, I decided to use a discrete-time event history model that includes time-varying covariates, with the risk onset of moving starting from the first month in residence spells. Some respondents who moved during the two year period (2013-14) have multiple residence spells available and my problem starts here. Previous study mostly goes with a multilevel modeling to account for correlation among multiple spells within the same person. However, in Stata, multilevel modeling cannot accommodate complex survey design using replicate weight.

Code:
svyset [pweight=wpfinwgt], vce(brr) brrweight(repwt1-repwt240) fay(.5) mse 
local varlist "timesince timesince2 timesince3 ib4.monthincomecat6 i.agecat i.educat i.hhtype childpres i.racecat" 
svy: melogit monthlymove `varlist' if insample == 1 || personid:
I would greatly appreciate any advice on using replicate weights in a multilevel model or any other suggestion on bypassing this issue.