intergenerational mobility that needs only cross-sectional data
and is based on the
informational content of surnames (????????????????????????????????????). Specifically, the ???????????????????????????????????? is defined as
???????????????????????????????????? ≡ ????????????
????????????
2
− ????????????
????????????
2
. The first term (????????????
????????????
2
) is obtained from the regression ????????????
????????????,????????????
= ????????????′???????????? +
???????????????????????????????????????????????????????????????????????????????????????????????? where ????????????
????????????,????????????
is the log of the income of individual ???????????? with the surname ???????????? and
???????????? is an S-vector of the surname-dummy variables with ????????????
????????????
= 1 if individual ???????????? has
the surname ???????????? and ????????????
????????????
= 0 otherwise. The second term (????????????
????????????
2
) is obtained from the
regression ????????????
????????????,????????????
= ????????????′???????????? + ???????????????????????????????????????????????????????????????????????????????????????????????? where ???????????? is an S-vector of “fake” dummy variables
that randomly assign surnames to individuals in a manner that maintains the
marginal distribution of surnames. The authors showed that the ICS is a
monotonically increasing function of the (more conventional) intergenerational
earnings elasticity and draws such a function for some baseline parameters.
Following this methodology, we estimate that earnings elasticity in the 15
th
century was between 0.8 and 0.9, thus depicting a quasi-immobile society. Then,
we compare these figures with those drawn from Güell et al. (2015a) for the
province of Florence in 2005 and analogously mapped into elasticity. These
findings are shown in Figure 7. Though they should be interpreted with some
caution, given the different nature of the data sources, they support the view that,
in the past, intergenerational mobility was (much) lower than it is today.
If one assumes that elasticities close to 1 were prevailing until the 20
th
century – i.e. before the effects of the industrial revolution were fully deployed in
Italy and before mass schooling – then one would obtain a long-run earnings
elasticity across six centuries that is comparable to ours.
6.2 Dynasties in elite professions
Our last empirical evidence concerns the existence of some degree of
persistence in certain (elite) professions. On the one hand, this represents a
further perspective (beyond earnings and wealth) on intergenerational mobility.
On the other hand, this evidence can provide some insight on the channels behind
intergenerational mobility processes.
Many social institutions contribute to status inheritance over multiple
generations, especially at the bottom (e.g. due to ethnic or social discrimination)
and at the top (e.g. membership of exclusive clubs and/or elite professions) of
hierarchies. In a society of perfect status inheritance (e.g. a pure caste system), the
children, parents, grandparents, and earlier ancestors are identical in their social
and economic positions; in this society, the perfect correlations between each
generation make alternative types of intergenerational effects (e.g. children-
parents, children-grandparents, etc.) indistinguishable. Zylberberg (2014)
21
underlined the existence of unobservable variables that are transmitted by
parents: The sons of successful families may preserve the high prospects for their
descendants, even when their own earnings are not very high. In his theoretical
framework, dynasties move across careers, rather than across income levels, and a
society can be modelled as a Markov process in which the transition matrix is
block-diagonal: only within-block mobility is allowed (e.g. the block of manual jobs
vs. the block of cognitive jobs). This is consistent with an earnings elasticity that
does not decline geometrically and with a society characterized by some form of
dynastic transmission of professions.
On the empirical side, we examine whether one’s probability of being
employed in a certain elite profession today is higher, the more one’s pseudo-
ancestors were employed in the same profession. Namely, we selected the
professions of lawyers, bankers, medical doctors and pharmacists, and goldsmiths.
We consider only these professions for several reasons. First, because of data
availability, we are forced to focus on professions that already existed in 1427 and
for which we currently have access to publicly available data. Second, they should
be elite or niche professions, consistent with the fact that there should be
unobservable variables that favored career following (e.g. specific human capital
or guild privileges). As shown in Figure 8, the earnings in the selected professions
are larger than the average, both in 1427 and today. Third, the available empirical
evidence documents the existence of career dynasties precisely for (some of) these
professions.
14
The results from the estimation of equation (3) are reported in Table 11. In
each column, we consider each profession separately, and we find a positive and
statistically significant correlation for lawyers, bankers and goldsmiths, and a
positive, but not significant, correlation for doctors and pharmacists. The
magnitude of the impact is clearly small. A one-standard deviation increase in the
independent variable increases the dependent variable by 0.5%, 0.2% and 0.6% of
its standard deviation for lawyers, bankers and goldsmiths, respectively.
Nevertheless, these results are, again, surprisingly high and strong if evaluated
across six centuries. Moreover, these results are consistent with earnings
persistence, and in particular, with larger persistence at the top of the earnings
distribution.
14
See Lentz and Laband (1989) for doctors, Laband and Lentz (1992) for lawyers and Mocetti
(2016) for pharmacists.
22