characterized by an unequal distribution across the families: the Gini
index was
nearly 40% for earnings and about 65% for real wealth. Members of the guilds
were at the top of the economic ladder and held influential positions in society and
politics. The most powerful guilds were those involved in the manufacture or trade
of wool and silk, and money changers. Indeed, many Florentine families were
successful bankers (e.g. Bardi, Medici and Peruzzi), and they were known
throughout Europe as well, for they established banking houses in other important
cities such as London, Geneva and Bruges.
Figure 1a groups the occupations, providing a complete picture of
occupational diversity and stratification. More than two fifths were artisans, such
as those who combed, carded and sorted wool, or carpenters. Entrepreneurs and
members of guilds, in turn, represented nearly one fifth of the workers. The
vibrant economic activity favored the development of lettered bureaucrats and
professionals (nearly one tenth of the workers) such as lawyers, judges, medical
doctors and pharmacists (the oldest pharmacy in Europe was set up in Florence).
Other significant occupational groups were those of merchants and of government
servants (e.g. firemen, town criers and soldiers). At the bottom of the occupational
ladder, there were unskilled workers, such as people beating, cleaning and
washing the raw wool, urban laborers and the servants of private families. When
mapping occupations into sectors of activity (Figure 1b), nearly half of the sample
was employed in manufacturing (mainly makers of wool and leather products).
Other important sectors were trade, food and wine, and public services. The
agriculture share, in contrast, was very small, because the data do not include the
countryside, which is where the agricultural activity was concentrated.
For slightly less than half of the surnames listed in the 1427 Census, we found
pseudo-descendants in the 2011 tax records. They correspond to about 800
surnames and 52,000 taxpayers. On average, they earn about 24,000 Euros per
year, and the real wealth is estimated to be larger than 160,000 Euros (Table 1).
Table 1 also shows that the professions under scrutiny are niche professions, both
in 1427 and in the 2000s: they account for a very small share of the workers.
Table 2 combines the tax records from 1427 and 2011 through surnames and
provides a first explorative assessment of persistence: we report for the top 5 and
bottom 5 earners among current taxpayers (at the surname level), the modal value
of the occupation and the percentiles in the earnings and wealth distribution in the
15
th
century (the surnames are replaced by capital letters for confidentiality
reasons). The top earners among the current taxpayers were already at the top of
the socioeconomic ladder 6 centuries ago: they were lawyers or members of the
wool, silk and shoemaker guilds; their earnings and wealth were always above the
13
median. On the contrary, the poorest surnames had less prestigious occupations,
and their earnings and wealth were below the median in most cases.
4. Main results
As shown in equation (1), in the first stage, we regress the log of the
ancestors’ earnings or the log of the ancestors’ real wealth on the surname
dummies (and, in some specifications, the controls included in the vector ????????????
????????????
????????????
) using
the 1427 Census data. We find that the surnames account for about 10% of the
total variation in the log of earnings and 17% of the total variation in the log of
wealth. These results support the hypothesis that the surnames carry information
about socioeconomic status. The coefficients for the surnames estimated in the
first stage are then used to predict the ancestors’ earnings and real wealth for the
taxpayers included in the 2011 tax records.
Table 3 presents our TS2SLS estimates of the intergenerational earnings
elasticity, as shown in equation (2). We consider three different empirical
specifications, with the first including only the predicted ancestors’ earnings, the
second and the third adding gender, and gender, age and its square, respectively.
The controls in the first stage regressions are adjusted accordingly. The earnings
elasticity is fairly stable across specification, with a magnitude around 0.04, and is
statistically significant at the 5% level. Table 3 also reports the standardized beta
coefficient and the rank-rank coefficient. According to the former, a one standard
deviation increase in the pseudo ancestors’ log earnings increases the pseudo
descendants’ log earnings by 6% of its standard deviation. The effect, besides being
positive and significant, is also non-negligible from an economic point of view. Put
differently, being the descendants of a family at the 90
th
percentile of earnings
distribution in 1427, instead of a family at the 10
th
percentile of the same
distribution, would entail a 5% increase in earnings among the current taxpayers.
Table 4 replicates the estimation with respect to the real wealth elasticity.
The parameter ranges from 0.02 to 0.03, and it is, again, highly significant. The
standardized beta coefficient equals 7% and is slightly larger than in the earnings
case. Stronger wealth persistence, with respect to earnings, is confirmed by the
results of the rank-rank regression (and similar findings are obtained, even if we
restrict the estimation to the same sample of families). The 10
th
-90
th
exercise
entails a 12% difference in real wealth today. The larger inertia in the real wealth
case is somewhat expected, as real wealth is accumulated through income (net of
consumption) over the life-cycle, but can also be directly passed down to
subsequent generations through bequests or inter-vivos transfers.
14