Koen Jochmans, "Testing for Correlation in Error-Component Models",
Journal of Applied Econometrics, Vol. 37, No. 7, 2020, pp. 860-878.

This paper uses the subsample of 12,360 mothers with three births from
the data set for the following paper:

  J. Abrevaya,"Estimating the effect of smoking on birth outcomes using
  a matched panel data approach", Journal of Applied Econometrics 21(4),
  2006, pp. 489-519.

The data are in the zipped ascii file birpanel.zip. The following
description of the data is taken from Abrevaya (2006):

momid3       identification number of the mother
idx          index number of a mother's birth (starts with 1 and goes
               to at least 2 for each mother); momid3 and idx together
               describe the panel structure of the data
stateres     code (1 through 51) for state of residence; see the NCHS
               (National Center for Health Statistics) website for codes
dmage        age of mother (in years)
dmeduc       education of mother (in years)
mplbir       state code (1 through 51) for mother's state of birth
nlbnl        number of live births now living
gestat       length of gestation (in weeks)
dbirwt       birthweight (in grams)
cigar        number of cigarettes smoked per day (99=unknown)
smoke        indicator variable for smoking status (1=smoker, 0=nonsmoker)
male         indicator variable for baby gender (1=male, 0=female)
year         year of birth (0=1990, ..., 8=1998)
married      indicator variable for marital status (1=married,
	       0=unmarried)
hsgrad       high-school graduate indicator (constructed from dmeduc)
somecoll     some-college indicator (constructed from dmeduc)
collgrad     college-graduate indicator (constructed from dmeduc)
agesq        age of mother squared
black        indicator variable for black race (1=black, 0=white)
adeqcode2    indicator that Kessner index = 2
adeqcode3    indicator that Kessner index = 3
novisit      indicator that no prenatal visit occurred
pretri2      indicator that first prenatal visit occurred in 2nd
	       trimester
pretri3      indicator that first prenatal visit occurred in 3nd
	       trimester
proxy_exists indicator variable for whether or not the proxy variable
	       for a correct match exists for this observation: if equal
	       to 1, the following variable contains the value of the
	       proxy; if equal to 0, the following variable contains
	       "proxyhat" (see below)
proxy_or_proxyhat  if proxy_exists = 1, this contains a proxy variable for
		   correct match (1 if interval since last live birth
		   matched the actual birthdate of the previous birth); if
		   proxy_exists = 0, this contains the predicted probability
		   that proxy variable is equal to one (see Section 4.2 of
		   the paper; this variable was constructed from a probit
		   regression using a mother's first and second births; the
		   probit specification included birthorder dummies,
		   education-level dummies, age, age squared, race, and
		   marital status)

To replicate the results in the paper in Stata, one can use the command
“xtserialpm”, which is described and illustrated in 

K. Jochmans and V. Verardi, "A portmanteau test for serial correlation
in a linear panel model", Stata Journal 20 (1), 2020, pp.149-161.

It can be downloaded from within Stata by typing "sic install
xtserialpm"