APPENDIX
Sources and criteria for statistical surveys followed by the Bank of Italy
A) SOURCES
A detailed description of the classification criteria can be found in "Supplementi al Bollettino
Statistico" , Banca d’Italia, Anno V Numero 32 - 16 June 1995. The dataset employed for this paper
has been provided by the Statistical Bureau of the Bank of Italy, on the basis of a specific request and
application. Therefore, although it has been structured and organized on the basis of our research plan, it
is necessarily based on the criteria of sampling and aggregation employed by the Bank of Italy for what
concerns the definitions of the different size classes of loans from different size classes of banks. In
addition, the degree of disaggregation is conditioned by the Italian regulation on banking secrecy, stating
that each disaggregated observation of credit flows from particular size classes of lenders, to particular
size classes of borrowers, for given geographic areas must be based at least on three observations. The
same applies to the data concerning interest rates, which are recorded by the Bank of Italy on the basis of
a non-compulsory quarterly report produced by a sample of 80 banks in all the country. For these reasons,
only a limited degree of disaggregation has been possible. The complete dataset employed for this
empirical analysis is contained in the folder “dataBI.zip” and cred.zip”, now available on the internet at
the site www.economia.unimo.it/worpap/ .
The file “dataBI.zip” contains the raw dataset obtained from the Bank of Italy in the file
“Modena.xls”, and the data processed for the estimates in all the other files of the folder contain. The file
“cred.zip” contain all the data processed for the descriptive analysis and the graphics.
The data on interbank and foreign deposits and securities held by banks in their portfolios are
based on reports from each individual bank to the Board of Inspection of the Bank of Italy. The data on
loan credit commitments, total used credit, total granted credit and bad debts are based on notice from
each individual bank to “Centrale dei Rischi” (the Department of Bank of Italy charged with monitoring
the system risk). The amounts are expressed in million lira and include both the operations in liras and in
foreign currency.
The interest rates on loan credits and on deposits have been obtained from the respective quarterly
statistical survey, which includes observations approximately 70 banks. The data on interest rates on bank
lending and deposits are only referred to the operations in liras.
B) LIMITS FOR THE STATISTICAL SURVEYS
From January 1996 the obligation to signal risks to “Centrale dei Rischi” only concerns loan cash
credits and credit commitments for amounts greater that 150 millions lira, while before that date the limit
was 80 millions lira. This creates a relevant discontinuity in the time series referred to the smallest loan
size, which, as a consequence, could not be employed for the econometric estimates and has been
included only from 1990 until 1996 in the graphics and descriptive statistics.
For the bad debts reported to Centrale dei rischi, there is no lower limit from february 1991, while
the limit was 10 millions lira before 1991. This creates yet another source of discontinuity in the time
series referred to the smallest loan size.
For what concerns the interest rates on loan cash credits and loan credit commitments, the lower
limits for the credit they are referred to is the same as the one for the credits reported to Centrale dei
Rischi. For the interest rates on deposits, the observations only refer to the deposits with amount larger
than 20 millions Italian lira. Loan cash credits until December 31st 1996 include overdue payments, while
from from January 1997 the latter are reported separately. This generates another time discontinuity for
all the loan sizes. data for net interbank deposits and foreign deposits include interbank transfers, while
any other variable variable in the dataset does not include them.
C) SAMPLING AND CLASS CRITERIA FOR THE VARIABLES OF THE DATASE
All the size classes have been calculated on the basis of loan cash credits granted by the whole
banking ssystem to each customer (see tables below).
28