Christopher H. Wheeler, "Evidence on Agglomeration Economies, Diseconomies,
and Growth", Journal of Applied Econometrics, Vol. 18, No. 1, 2003,
79-104.

There are two data files:  one for the metropolitan areas (chwmetro.data)
and the other for the counties (chwcnty.data).  Both are comma delimited.
Variables appear in columns in both cases with variable names in the 
first row.

Both files, which are in DOS format, are zipped in the file
wheeler-data.zip. Unix users should use "unzip -a".

Metropolitan areas are defined as metropolitan statistical areas (MSAs)
or consolidated MSAs (CMSAs) if an MSA belongs to a CMSA.  Employment and 
population figures are calculated by summing all county-level observations
within each metropolitan area.  MSA and CMSA definitions from the year 
1990 are used.  Definitions appear in the USA Counties data set, listed 
below.

Data Source:  Bureau of the Census.  (1999) USA Counties 1998 on CD-ROM.
        [machine readable data file] The Bureau: Washington, DC.


1.  Data file chwmetro.data

        Observations: 268 MSAs/CMSAs

        Variables appear in the following order: 
        
        METRO = metropolitan area code (1990 definitions)
        EMPGR = employment growth rate (1980-90)
        POPGR = population growth rate (1980-90)
        LOGEMP = natural logarithm of employment 1980
        LOGPOP = natural logarithm of population 1980


2.  Data file chwcnty.data

        Observations: 3106 counties (3102 when using EDUCSH, HWSH, POLSH)

        Variables appear in the following order:
        
        STATE = state FIPS code
        COUNTY = county FIPS code
        EMPGR = employment growth rate (1980-90)
        POPGR = population growth rate (1980-90)
        DEMPGR = employment growth rate minus 
                 mean across all counties in same state (1980-90)
        DPOPGR = population growth rate minus 
                 mean across all counties in same state (1980-90)
        LOGEMP = natural logarithm of employment 1980
        LOGPOP = natural logarithm of population 1980
        EDENS = employment density (employment per square mile) 1980
        PDENS = population density (population per square mile) 1980
        LOGAREA = natural logarithm of land area 1980
        COLLRATE = fraction of adult population with bachelor's degree 
                or more 1980
        MFGRATE = fraction of employment in manufacturing 1980
        UR = unemployment rate 1980
        PCINC = per capita income (dollars) 1979
        EDUCSH = share of local government spending on education 1982
        HWSH = share of local government spending on highways 1982
        POLSH = share of local government spending on police 1982
        NWRATE = fraction of population that is not white 1980
        URBAN = urban indicator variable 
                (= 1 if county located in metropolitan area) 1990
        RURAL = rural indicator variable 
                ( = 1 if county located outside metropolitan area) 1990