Gary Koop and Justin L. Tobias, "Learning about Heterogeneity in Returns
to Schooling", Journal of Applied Econometrics, Vol. 19, No. 7, 2004,
pp. 827-849.

This panel data set consists of NT=17,919 observations from 
N=2,178 individuals. The data are taken from the National 
Longitudinal Survey of Youth. 

Both data files are ASCII files in DOS format. They are zipped in the file
kt-data.zip. Unix users should use "unzip -a".

The data set is broken into two parts. The first part, "time_var.dat",
contains the time-varying characteristics together with the
individual-identification vector (denoted person_id). This file contains
17,919 observations on 5 variables. These variables are: 

  Column 1: Person_id (Ranging from 1-2,178). 
  Column 2: Education 
  Column 3: Log Hourly Wage 
  Column 4: Potential Experience 
  Column 5: Time Trend 

(NOTE: The paper also uses a continuous local unemployment 
rate variable, but this variable is not included here 
as it is part of the restricted GEOCODE version of the NLSY). 

The second part, "time_invar.dat," contains the time-invariant variables.
It contains 2,178 observations on 5 variables. These are:

  Column 1: Ability 
  Column 2: Mother's Education
  Column 3: Father's Education
  Column 4: Dummy for Residence in Broken Home 
  Column 5: Number of Siblings 

NOTE 1: The data sets are arranged so that the individual from
"time_var.dat" with person_id=j has the time-invariant characteristics
listed in the j^{th} row of "time_invar.dat".

NOTE 2: When performing analyses in the paper with the inclusion of the
time_invariant characteristics, we excluded observations with parental
education being less than 9.