I am estimating a model with 1056 observations over 586 days(with gaps on
days with no transactions) of transactions in a commodity market by about
300 customers (i.e., i is 300, t = 586). Some customers are observed over 50
times; others appear only once. In addition, on some days no transactions
occur so the 586 days of transaction data actually spans 2 years.
I would appreciate if someone could let me know whether STATA's xtreg (both
fe and re versions) and related xt estimation commands, are suitable for
estimating unbalanced panels, or whether one needs to make adjustments to
the estimated parameters and standard errors for the highly unbalanced
nature of my panel (I could balance it by adding a ton of zeros for all the
non-transacting customers every day, but that doesn't sound as if its adding
any real information to the system.
