Tuscany 1427
Income Class

Number of people

Percentage of people

Average income per capita (florins per annum)

Income in terms of per capita mean

Lowest decile

3918

10.2

10.90

0.32

Second

3936

10.3

12.99

0.38

Third

3659

9.6

14.59

0.43

Fourth

3937

10.3

16.60

0.48

Fifth

3776

9.9

18.99

0.55

Sixth

3841

10.0

22.26

0.65

Seventh

3750

9.8

26.47

0.77

Eights

3864

10.1

33.44

0.98

Ninth

3796

9.9

47.98

1.40

Top decile

3827

10.0

138.79

4.05

Total

38304

100

34.23

1

Income distribution data: The underlying data from the special census of 14271429 cover the city of Florence and immediate Tuscan environs, and not the other territories that Florence controlled between the Appenines and the Mediterranean, using data from 1427 only. The data were originally collated in a famous study by David Herlihy and Christiane KlapischSupan (1985). A newer version consists of the downloadable files at the Brown University site http://www.stg.brown.edu / projects / catasto. Peter Lindert later (January 2008) downloaded the same source’s data on bocche (persons) and on real estate. We thank Maristella Botticini of Boston University for downloading most of the variables.
The assumptions behind our estimates are inevitably complex, because the catasto return itself is complex.
Herlihy and KlapischSupan (HK) capitalized yearly income estimates to get wealth at the rate of 7 percent. Thus, by taking 7% of wealth as property income, we are reversing their procedure.
Those assets exempted from the assessment included “the family home and its furnishings, and also the tools which supported the taxpayer in productive employment. They [also] gave favorable treatment to plow animals in the countryside and beasts of burden everywhere. They even allowed deductions for the costs of maintaining farm buildings and fertilizing fields.” (Herlihy and KlapischSupan 1985, pp. 910).
More exemptions: “[E]very citizen of Florence was allowed to subtract 200 florins from his total assets for every family member.” (p. 10.) We assume that this did not reduce the stated wealth figures here, and only reduced the taxes paid.
But “no deductions [of 200 florins of wealth, or 14 florins of income] be allowed for the mouths or heads of any salaried person, whether servant, nurse, clerk, employee or apprentice.” (Source, as quoted in Herlihy and KlapischSupan 1985, p. 12.)
“By far the largest group of exempt persons was the clergy. From the number of parishes and religious institutions in Tuscany, we would estimate their size at some 7,000 to 8,000” (p. 25.) But this seems to be for all the territories controlled by Florence, not just for the city.
The assessments are apparently of wealth and income, not just taxable wealth and income.
We made three alternative sets of assumptions about income inequality in Florence 1427: TOOEQUAL = A set of assumptions that with high probability will understate the inequality of income among households (and individuals); PREFERRED = A set of assumptions designed to estimate the medianprobability Gini coefficient for incomes; and TOOUNEQ = A set of assumptions that with high probability will overstate the inequality of income. [In addition, ALL = An assumption common to all three estimates. Such an assumption cannot, of course, be both tooequal and toounequal at the same time, but it is our belief that the other extreme assumptions succeed in putting bounds on the Gini coefficient.]
Asset incomes:
(1) We accept the Catasto’s wealth estimates and its 7% rate of return as accurate. Exceptions are a few cases where the census data on the existence of an asset conflict with its zero valuations, as noted below.
Labor earnings rates:
(2) We matched 26 laborintensive occupations from the catasto with four skills groups yielding direct wage estimates from other sources. We assumed that ordinary workers earned wages or salaries 260 days a year. For the other, more propertied, classes, we assumed: (2a) The TOOEQUAL ASSUMPTION: Nobody had a labor income per earner that was higher than the 66 florins a year earned by the average clerk working on the catasto. This seriously underrates the current earnings of managerial and highly skilled labor. (2b) The PREFERRED ASSUMPTION: For these more propertied classes, labor income = the midpoint between the labor income estimates in (3a) and (3c) below. (2c) TOOUNEQUAL ASSUMPTION: Labor income = max (14 florins, the value of property income). This assumption denies the fact that incomes from nonhuman assets like land, bonds, and even commercial property must have soared above the labor earnings in the top quarter of the distribution.
(3) Earnings by nonheads in the household: Charles de la Roncière assumes that the expenditures (and, implicitly, income) of a Florentine worker’s family of four people in 13691377 was 2.22 times the earnings of a bachelor with the same occupation.
So we assume that each nonhead member of a workingclass household raised income by 0.407 (=1.22/3) times the unskilled wage rate of 23.1 florins, or 9.4 florins. Such additional labor earnings were probably greater in lowerstatus households than in more propertied households, but we cannot assume so. Thus: (3a) TOOEQUAL in TOTAL INCOME: Apply these nonhead earnings only to the laborintensive stated occupations. (3b) PREFERRED: Apply them to the stated laborintensive occupations and to nonstated occupations, not to highstatus. (3c) TOOUNEQ: Apply these extra earnings to all households.
For a tooequal distribution for income per capita we added (3d) TOOEQUAL in PER CAPITA: Assume that each nonhead in the households with unskilled wage with stated laborintensive occupations earned 4/5 of the unskilled wage, or 18.48 florins.
Home ownership, for those with occupations given:
(4) ALL: (4a) No adjustment is needed for the income value of owneroccupied housing in cases where the home’s value was assessed. This is because the surveyors included such implicit income in their valuations.
We need a different approach for families where no realestate value is given.
(4b) Where no realestate value is given and the household does not own its residence (this should be a redundancy, but isn’t one in fact), again no adjustment is needed.
(4c) BUT in cases where there is no realestate value but the household is recorded as owning its home, we must multiply the stated incomes (property plus assumed labor earnings) by 1/(1ρ).
Rho is the share of rental expenditures in total expenditures among households in Florence in 13691377 according to de la Roncière. It equals 3.8 percent for bachelors and 6.8 percent for a family of four. We use the 6.8 percent figure for all cases where bocche (persons) ≥ 2.
(4d) Similarly, in 15 percent of cases where the household head either gets housing rentfree or nothing is said about ownership or rent, we multiply by 1/(1ρ).
Rural households (Herlihy and KlapischSupan consider the nooccupation to be heavily rural):
(5a) TOOEQUAL and PREFERRED estimates: No occupation listed, no cattle, and no home ownership, assume labor earnings = 14 for main earner.
(5b) ALL: No occupation, no cattle, owns home, and has real estate: assume an unskilled wage of 23.1.
(5c) ALL: No occupation, no cattle, owns home, BUT has no recorded real estate: again assume an unskilled wage of 23.1, but in these cases we must multiply total income by 1/(1rho) because of the implicit value of housing, as in (5c) above.
(6) ALL: No occupation, no home ownership, but owns some beasts or cattle: A pretty rich group. Use 23.1 for the household head’s labor income.
(7) No occupation, yes home owned, and owned some beasts and cattle (these tended to have aboveaverage property income):
(7a) Head’s income for TOOEQUAL = 23.1.
(7b) Head’s total income for PREFERRED = 23.1 * 1.5.
(7c) Head’s total income for TOOUNEQ = 23.1 * 2.
(8) TOOUNEQUAL: As with the occupationstated group,
Labor income = max (14 florins, and 1/4 the value of property income).
This overstates rural inequality, while honoring the likelihood that labor incomes were a much smaller share of high rural incomes than of high urban incomes.
Gini estimates:
Tuscan Gini coefficients (Gini1 = Gini2) for 1427, estimated using the assumptions described above:

Tooequal
(underestimate)

Preferred estimate

Toounequal
(overestimate)

Gini for total household incomes

53.0

54.9

59.9

Gini for household income per capita

44.0

47.1

50.5

Population and area: Population taken directly from the census. Total number of households in the census is 9,779, and total population (bocche) is 38,340. Area from Herlihy and KlapischZuber (1985, p.39).
Urbanization rate: From Herlihy and KlapischZuber (1985, p. 56).
Mean income in $PPP: We took the ratio of average income per capita for Tuscany (34.6 florins) to its subsistence income estimated at 14 florins a year, and applied this ratio (2.47) to the assumed level of subsistence of $PPP300. Back in 1301 “Florence’s yearly income was then an estimated 780,000 pounds or roughly 338,000 florins” (Herlihy and KlapischSupan, p. 2). Thus our estimate of 1,287,607 florins for 1427 implies a growth rate of 1.06 percent a year for the combination of average real incomes, population, and price inflation.
REFERENCES
http://www.stg.brown.edu/projects/catasto/overview.html, accessed January 2008. Edited by David Herlihy, Christiane KlapischZuber, R. Burr Litchfield and Anthony Molho. The Online Catasto is a World Wide Web searchable database of tax information for the city of Florence in 142729 (c. 10,000 records). It is based on David Herlihy and Christiane KlapischZuber, Principal Investigators, “Census and Property Survey of Florentine Dominions in the Province of Tuscany, 14271480.”
Cipolla, Carlo M. (1948), Studi di storia della moneta. Pavia: Università de Pavia.
de la Roncière, Charles (1982), Prix et Salaires à Florence au XIVe Siècle, 12801380. Palais Farnèse: Ecole Française de Rome, pp. 394395.
Herlihy, David, and Christiane KlapischZuber (1985), Tuscans and their Families. New Haven: Yale University Press.
Goldthwaite, Richard A. (1980), The Building of Renaissance Florence. Baltimore: The Johns Hopkins University Press.
Malanima, Paolo (2002), L’economia italiana: dalla crescita medievale alla crescita contemporanea. Bologna: Società editrice il Mulino.
