Table 4. Real wealth mobility: baseline
Dependent variable:
Log of wealth Log of wealth Log of wealth
Log of ancestors’ wealth
0.027***
0.026***
0.018**
Standardized beta coefficient
0.103
0.102
0.069
(0.008)
(0.008)
(0.008)
Rank-rank coefficient
0.105***
0.105***
0.073**
(0.031)
(0.031)
(0.030)
Female
NO
YES
YES
Age
and age squared
NO
NO
YES
Observations
679
679
679
R-squared
0.018
0.020
0.110
Bootstrapped standard errors in parentheses (1,000 replications); *** p<0.01, ** p<0.05, * p<0.1.
Table 5. Earnings mobility: transition matrix
Origin ↓ / Destination→
Lower
class
Middle class
Upper
class
Lower class
32.8
36.4
30.8
Middle class
43.0
29.1
27.9
Upper class
25.3
34.8
39.9
Table 6. Real wealth mobility: transition matrix
Origin ↓ / Destination→
Lower
class
Middle class
Upper
class
Lower class
41.6
29.8
28.6
Middle class
31.6
34.3
34.1
Upper class
26.8
36.2
37.0
29
Table 7. Earnings and real wealth mobility: robustness
Dependent variable:
Log of earnings
Log of wealth
Log of ancestors’ earnings/wealth
0.065*
0.061**
0.023**
0.016**
(0.033)
(0.030)
(0.010)
(0.008)
Controls
YES
YES
YES
YES
Specification
Imputation
procedure
Trimming Imputation
procedure
Trimming
Observations
806
790
679
667
R-squared
0.068
0.048
0.085
0.101
In column 1, earnings are corrected using the parameters estimated by Marino and Zizza (2011); in column 3, we use real estate incomes instead
of real wealth imputed through the SHIW; columns 2 and 4 refer to the exclusion of the top and bottom percentile of both the dependent and
independent variables. Controls include a dummy for female and age and age squared. Bootstrapped standard errors in parentheses (1,000
replications); *** p<0.01, ** p<0.05, * p<0.1.
Table 8. Mobility for rare and Florence-specific surnames
Dependent variable:
Log of
earnings
Log of
wealth
Log of
earnings
Log of
wealth
Log of ancestors’ earnings/wealth
0.076**
0.019**
(0.034)
(0.009)
×
Less typical Florentine surnames
0.021
0.014
(0.036)
(0.010)
×
More typical Florentine surnames
0.053*
0.020*
(0.027)
(0.011)
Controls
YES
YES
YES
YES
Specification
More weight to
rare surnames in 1427
Differences by low- high-
Florence-specific surnames
Observations
806
679
806
679
R-squared
0.061
0.119
0.049
0.110
More (less) typical Florentine surnames are those for which the ratio between the surname share in Florence and the corresponding figure at the
national level is above (below) the median. Controls include a dummy for female and age and age squared. Bootstrapped standard errors in
parentheses (1,000 replications); *** p<0.01, ** p<0.05, * p<0.1.
30
Table 9. Earnings and wealth distribution by survival rate
Surviving
families
Missing
families
Difference
Log of ancestors’ earnings
3.465
3.406
0.059** (0.026)
Log of ancestors’ wealth
4.628
4.504
0.124 (0.115)
Surviving families refer to surnames that are present both in 1427 Census and in 2011 tax records; missing
families are surnames existing in 1427 Census but not in 2011 tax records; standard errors in parenthesis; ***
p<0.01, ** p<0.05, * p<0.1.
Table 10. Heckman corrected estimates
Dependent variable:
Log of earnings
Log of wealth
Log of ancestors’ earnings/wealth
0.047*
0.025***
(0.027)
(0.009)
Controls
YES
YES
Inverse Mills’ ratio
0.008
0.226*
(0.042)
(0.130)
Observations
806
679
Probability of surviving
Size of the family in 1427
0.008***
0.003***
(0.001)
(0.000)
Observations
1,895
1,895
Controls include a dummy for female and age and age squared. Bootstrapped standard errors in parentheses (1,000
replications); *** p<0.01, ** p<0.05, * p<0.1.
Table 11. Probability to belong to a given profession
Dependent variable:
Lawyer
Banker
Doctor or
pharmacist
Goldsmith
Share of ancestors in the same profession
0.004*** 0.001**
0.001
0.004***
(0.001)
(0.000)
(0.002)
(0.001)
Observations
133,193 133,193
133,193
133,193
R-squared
0.000
0.000
0.000
0.000
Marginal effects from a probit model are reported. Standard errors clustered at the
surname level in parentheses; *** p<0.01, ** p<0.05, *
p<0.1.
31