Let us first address information on yearly transitions for owners of small business in period t-
1 and period t. Table 4 presents mobility tables, based on quintiles, for three periods, 1993-1994,
1997-1998, and 2002-2003, and for the two definitions. Results are presented in terms of
(unconditional) percentages in each cell, showing the percentage of owners of small businesses
combining being in quintile r in period t-1 and quintile s in period t. For instance, Table 4 shows that
of all the owners of small businesses, according to the narrow definition, 21.6 of them were in quintile
1 both at time t-1 and at time t.
Upward mobility will be reflected by a larger percentage in higher quintiles in period t,
compared to period t-1, see Table 4. We also expect the wider definition to show higher upward
mobility than the narrow definition, probably towards the end of the period, i.e., 1997-1998 and 2002-
2003 compared to 1993-94.
Table 4 does not indicate substantial upward mobility of owners of small businesses. The
share of owners of small firms in quintile 5 in year t-1 is little different from the share in year t;
compare column totals and row totals for quintile 5. As expected, the wider definition results in better
performance for owners of small businesses, but we do not observe increased upwards movement in
the period. In fact, there were fewer owners in quintile 5 in 2003 than in 2002. If we extend the
income measurement period, i.e., transitions when rankings (in both t-1 and t) are determined by
aggregate post-tax income for several years (aggregate incomes for period 1993-1998 and period
1998-2003), the pattern (not shown) is basically the same as seen in Table 4. Similar results are also
seen for a longer transition period, e.g., comparing transitions between 1993 (t-1) and 2003 (t).
Such tables can straightforwardly be turned into indices of overall mobility. Let us therefore
first consider what estimates of indices of mobility say about overall mobility of business owners and
wage earners. Table 5 presents estimates of overall mobility over the three periods for the three
subgroups, for two measures of income mobility. The two mobility measures are defined in terms of
conditional transition probabilities, pjk , the probability that an individual moves to class k given that
he/she was initially in class j. The mapping of the probabilities into indices of mobility follows
suggestions by Prais (1955)/Shorrocks (1978) (PS) and Bartholomew (1982) (B), seen as
m
m-∑pjj
(1) ps =---j=L_
m-1
and
mm
(2) B = mrl ∑∑ ∏jPjkj k∣∙
m- j=1 k=1
14