located grouped data. These data comprise age (3 groups), sex (2), educational qualification
(5), industry (132), year (4) (to remove the effect of inflation on incomes). This gives
1,034,000 cells for each of the 4 years (275 x 3 x 2 x 5 x 132), in all 4,356.000 cells. 20,8% of
these cells have non zero content with an average of 9.9 employees per observation In
addition, there is a variable containing unemployment percentages for each age by sex by
qualification by year for each municipality category (275 x 3 x 2 x 5 = 8.250 for each of the 4
years 33,000 possible different values). Finally, there is a distance variable, based upon
location of the municipality, which has 275 possible different values. These distances are
expressed as monetary values. Three distance measures are used: i) to the urban centre in the
(statistically defined) Danish labour market areas, ii) to the capital Copenhagen, and iii) to the
nearest of the five large university towns: Copenhagen, Ârhus, Odense, Aalborg, and Esbjerg.
The dependent variable is mean value of wages and salaries per person (Full-time equivalent)
for each of the 66.000 cells for each of the 4 years.
Data used to estimate the pecuniary externality effects has the following structure. The
dependent variable is mean value of wages and salaries per person defined in relation to the
categories used to define the values of the independent variables. Grouped data was used as
data at individual level was available. The independent variables are principally category
variables, representing, for each combination of the 275 municipalities where place of
production is located and each of the same municipalities where place of residence is located.
In the data set used to estimate pecuniary externalities, it should be noted that basic data
relates to groups of individual data by place of residence. These data comprise age (3 groups),
sex (2), qualification (5), year (4) (to remove the effect of inflation on incomes). This gives
2,268,750 cells for each of the 4 years (275 x 275 x 3 x 2 x 5), in all 9,075,000 cells over the 4
years. This contrast with data used to estimate urban externalities, where the basic data refers
to individuals grouped by industry and there is no information concerning place of residence
in this data set. In addition, there is a variable containing unemployment percentages by place
of residence for each age by sex by qualification by year category (275 x 3 x 2 x 5 = 8250
cells for each of the 4 years. Finally, there is a distance variable, based upon the 275 x 275
inter-municipality distance matrix. The dependent variable is mean value of wages and
salaries per person (Full-time equivalent) for each of the 2,268,750 observations per year (=
275 x 275 x 3 x 2 x 5).
The number of observations is of course much lower than the number of cells: Firstly,
especially in the data set on pecuniary externalities the majority of cells are empty. Second, a
number of observations have been eliminated for different reasons.
The following observations have eliminated from the data set used for the analysis of urban
externalities: If place of work is abroad, not available, or located on the small island of
Christians0, and if age is below 15 or above 59. Furthermore, some extreme observations are
removed, if the average wage is above 2 mill. DKK, and if total employment is under 5
working days the year concerned. This all reduces the number of observations to 277,237 per
year in the 4-year-period with an average of 9.0 employees per observation and a standard
deviation of about 35 workers. About 1 per cent of the observations contain more than 100
employees. The highest number of employees in an observation is around 2,200. Similar
reductions have been made for the data set used to analyse pecuniary externalities. The total
number of observations here is 205,680 per year with an average of 9.9 employees per
observation and a larger standard deviation of about 89 employees.
Different measures of distance are applied in the regressions, but the source is the same.
Distances are at municipal level. The distance from one municipality to another is the number
of kilometres between the main post office in every pair of municipalities. An intra-
municipality distance is calculated including elements such as size and shape of the