17
at 100.000), in view of the higher Swedisch income level,
conceals the relevant part of the Pareto distribution
which presumably would show a much higher Pareto co
efficient than the income below 100.000 crowns does.
For 1958/59 (see / /) data are available on the various
types of income ofjproperty owners in Holland:1) income
of unincorporated business and professions 2) unearned
income, and 3) other income (which is chiefly income
from employment). The simple regressions of these three
incomes on the total income of property owners have been
calculated (fig. 4). The regression coefficients are
respectively b^ = 0.80, l>2 - 1.34 and b^ = 1.01.
This shows that with increasing wealth the share of
income derived from property is increasing.
The Pareto coefficient for wealth in 1958/59 is 1.57»
for income of property owners it is 2.08. If we estimate
it on the basis of the theory, the coefficient for in
come would be 1.94 (since k in that year is about 0,81).
On the basis of the above information we can derive
theoretical estimates of Pareto coefficients for the above
mentioned three income types separately from the Pareto
coefficient for total income of property owners, which in
turn can be derived from the coefficient for the wealth
distribution. We can write for the density of, for