I am trying to replicate what's called a 'dynamic autoregressive' logit model to predict a binary outcome variable. So the model looks like: Array


So how would I go about lagging vt so that I can estimate a logit model that also has vt-1 as a predictor? Like such: Array


My code so far looks like this (where USREC is my binary outcome):
Code:
ssc install freduse
freduse USREC UNRATE CMRMTSPL FEDFUNDS INDPRO
gen t = mofd(daten)
format t %tm
tsset t
keep if tin(1969m1, 2020m2)


then modelling a logit without the lag of v would look like this (and let's say, for simplicity, I use 5 lags):

Code:
logit USREC L1.USREC L(1/5).D.UNRATE L(1/5).D.CMRMTSPL L(1/5).D.FEDFUNDS L(1/5).D.INDPRO if tin(1969m1,2019m2)
So how would I lag the entire logit equation (to match the third equation above) so I can estimate another logit model with the included lag of itself? Thanks.