cover the whole population, they are rich in information. 1993 was chosen as the first year of the
analysis because that was when the register itself was established.
In terms of scope we restrict the analysis to individuals with a connection to the workforce;
pensioners and students are therefore excluded. We did this by excluding all except prime aged
individuals, i.e., persons aged 25-55 in the period 1993-2003, earning a wage or business income in
excess of the Basic Pension Unit of the National Insurance Scheme in every year of the period. In
1993 the Basic Pension Unit was NOK 37,033 (USD 5,230), rising to NOK 55,964 (USD 7,900) in
2003. After these constraints, we were left with a sample of approximately 900,000 individuals, wage
earners and self-employed. As data come from public records, the present analysis does not suffer
from attrition problems, as frequently reported in other panel data analyses.
Farmers and fishermen are usually defined as self-employed.5 However, as Norwegian
primary industries are heavily subsidized and regulated, we excluded these groups from the sample.
Nor are they included in the categories “owners of small businesses” or “wage earners”.
As already noted, there are three categories of small business owner: self-employed; owners
of closely held corporations under the split-model; and owners of widely held corporations. The self-
employed are identified by reporting business income from self-employment (positive or negative).
Owners of closely held corporations are recognized in that they report imputed labor income under the
split model in combination with reporting wage income.6 The main problem, when addressing this
issue empirically, is that it has previously been impossible to identify owners of small firms who run
their business as a widely held firm. Income tax return data let us observe individuals who combine
wage earning status with status as a recipient of dividends, but so far we have been unable to ascertain
whether the wages and dividends come from the same firm. Obviously, this is essential for identifying
owners of small businesses. However, the new Register of Shareholders has changed all this. By
combining information from the Register of Shareholders and wage information from the End of the
Year Certificate Register, which tells us which firm is paying the individual’s wages, it is now
possible to find wage earners who are simultaneously employees and major shareholders of the same
firm. However, 2004 was the first year for which the Register of Shareholders issued data, so we rely
on data from 2004 to establish end-of-period status only. As we have a longitudinal perspective in this
analysis, we need to make assumptions about status in previous years. We do this as follows. We let
information on status in 2004 determine status in 2003. If subjects, according to the data, were likely
wage earners before 2003, we assume it holds true for the preceding years. If they are identified as
self-employed or owners of a closely held corporation in preceding years, the business will likely have
5 See Parker (2004) for more details on different definitions of firms.