The Role of Social Capital in Financial Development

The Role of Social Capital in Financial Development
Author(s): Luigi Guiso, Paola Sapienza and Luigi Zingales

Source: The American Economic Review , Jun., 2004, Vol. 94, No. 3 (Jun., 2004), pp. 526-
556
Published by: American Economic Association
Stable URL: https://www.jstor.org/stable/3592941
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
is collaborating with JSTOR to digitize, preserve and extend access to The American Economic
Review
This content downloaded from

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
All use subject to https://about.jstor.org/terms
The Role of Social Capital in Financial Development
By LUIGI GUISO, PAOLA SAPIENZA, AND LUIGI ZINGALES*
To identify the effect of social capital on financial development, we exploit social

capital differences within Italy. In high-social-capital areas, households are more
likely to use checks, invest less in cash and more in stock, have higher access to
institutional credit, and make less use of informal credit. The effect of social capital
is stronger where legal enforcement is weaker and among less educated people.
These results are not driven by omitted environmental variables, since we show that
the behavior of movers is still affected by the level of social capital of the province
where they were born. (JEL Z13, G10, 016)
In 1958 when Edward C. Banfield wrote developments in economic theory allow us to

"The Moral Basis of a Backward Society" few the intrinsic limitations agents face
appreciate
economists, with the exception of Kenneth Ar-
in contracting and the potential role social cap-
ital, and the trust it engenders, can play in
row, noticed.' His thesis that the underdevelop-
ment of southern Italy was due to the lack of the deadweight loss generated by these
reducing
social trust outside the strict familylimitations.
circle
For this reason, the work of Robert D. Put-
(which he labeled "amoral familism") was hard
nam (1993) and Francis Fukuyama (1995) has
to reconcile with the economic models prevail-
captured the attention of several economists.
ing at that time. Forty years later, however,
Rafael La Porta et al. (1997a), for example,
document a strong correlation between the trust
prevailing
* Guiso: Department of Economics, University of Sas- in a country and the presence of large
organizations. Similarly, Stephen Knack and
sari, Ente L. Einaudi for Monetary Banking and Financial
Studies, Via Due Macelli 73, Rome 00187, Italy, and CEPR
Philip Keefer (1997) find a correlation between
(e-mail: [email protected]); Sapienza: Kellogg School of Man-
a country's level of trust and its rate of growth.
agement, Northwestern University, Evanston, IL 60208, and
This
CEPR (e-mail: [email protected]); Zin- correlation persists even after controlling
gales: Graduate School of Business, University of for quality of law enforcement (Knack and Paul
Chicago,
1101 East 58th Street, Chicago, IL 60637, NBER, Zack,
and2001).
CEPR (e-mail: [email protected]). We thank
The skeptics, however, could still object.
three anonymous referees, Sonia Falconieri, Bruno Gerard,
First, people's
Kathleen Johnson, Tarun Khanna, Harris H. Kim, Arvind
trust may be the result not only
of the social
Krishnamurthy, Owen Lamont, Elizabeth Paparo, Giancarlo capital present in their community,
Spagnolo, Peter Tufano, Brian Uzzi, and Daniel but also of
Wolfen- prompt law enforcement. Second,
zohn for very helpful comments. We also benefited the
fromtheoretical
the link between social capital and
comments of participants of seminars at CORE, New York
growth is very indirect (e.g., Robert M. Solow,
University, Northwestern University, CEMFI, Stanford
1995). Even Putnam (1993) admits that the
University, University College, the Finance Brown Bag
lunch at the University of Chicago, University ofmechanisms
Florida, through which "the norms and net-
University of Maastricht, 2000 WFA Conference, work
NBER of the civic community contribute to eco-
Corporate Finance Meeting, and BIRC 5th Scientific Coun-
nomic prosperity" should be investigated
cil Meeting. Luigi Guiso acknowledges financialfurther.
support
from MURST, and the EEC under the RTN program "Spe-
In this paper we investigate the link between
cialization versus Diversification." Luigi Zingales acknowl-
edges financial support from the Center for Researchthe level
on of social capital and one important
Security Prices at the University of Chicago. We thank
factor underlying economic prosperity, financial
Sandra Sizer Moore for editorial assistance.
development. One of the mechanisms through
1 Arrow (1972) wrote: "It can be plausibly argued that
much of the economic backwardness in the world can be which social capital impacts economic effi-
explained by the lack of mutual confidence. See Banfield'sciency is by enhancing the prevailing level of
remarkable study of a small community in southern Italy."trust. In high-social-capital communities, peo-
526

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
VOL. 94 NO. 3 GUISO ET AL.: SOCIAL CAPITAL IN FINANCIAL DEVELOPMENT 527
are neither legal nor economic incentives to

pie may trust each other more because the net-
works in their community provide better donate blood or to vote. Both decisions are
opportunities to punish deviants (James driven
S. only by social pressure and internal
Coleman, 1990; Giancarlo Spagnolo, 1999). At i.e., the fundamental components of so-
norms,
the same time, in these communities people
cial capital.
We study the effect of social capital on a
may rely more on others' keeping their prom-
ises because of the moral attitude imprinted
variety of households' financial choices: use of
with education (Banfield, 1958). checks, portfolio allocation, availability of
Since financial contracts are the ultimate loans, and reliance on informal lending. Con-
trust-intensive contracts, social capital shouldsistent with social capital being important, we
have major effects on the development of finan- find that in areas characterized by high levels of
cial markets. Financing is nothing but an ex- social capital, households invest a smaller pro-
change of a sum of money today for a promise portion of their financial wealth in cash and a
to return more money in the future. Whether bigger proportion in stock. This result holds
such an exchange can take place depends not even after we control for a large set of house-
holds' characteristics and some other environ-
only on the legal enforceability of contracts, but
also on the extent to which the financier trusts mental variables, such as quality of legal
the financee. enforcement, the average level of education,
Since social capital is an important determi- and the per capita gross domestic product
nant of the level of trust, it should also affect the (GDP). In social-capital-intensive areas, house-
level of financial development. Documenting holds are also more likely to use personal
this link can not only shed light on the mecha- checks and to obtain credit when they demand
nism through which social capital contributes to it. These results are not driven by omitted en-
economic prosperity, but also provide a new vironmental variables, since we show that the
explanation for the widely different levels of behavior of those people who move is still af-
financial development across countries. fected by the level of social capital present in
The use and availability of financial contracts the province in which they were born.
across countries is affected by many other in- We find that the likelihood of receiving a loan
stitutional factors, and thus is difficult to con- from a relative or a close friend decreases with
trol for in a regression. Therefore, we exploit the level of social capital that prevails in the
within-country variations to identify the effects area. This finding is consistent with Banfield's
of social capital on the use and availability of (1958) and Fukuyama's (1995) claims that
financial contracts. We use Italy as our sample low-social-capital areas are often character-
country, both for the availability of detailed ized by more intense reliance on transactions
microeconomic data and for its historical im- within narrow subgroups, such as families
portance in the social capital debate: Italy is theand friends.
country where sociologists first turned to study To examine the causal nature of these corre-
the effects of trust and social capital (Banfield, lations, we explore whether the magnitude of
1958; Putnam, 1993). the impact of social capital varies as theory
The most contentious issue is how to measure predicts. Consistent with theory, we find that the
social capital. Since the concept itself is com- effect of social capital is stronger when legal
plex, most of the measures used in the literatureenforcement is weaker. We also find that the
are outcome based, e.g., the level of trust oreffect of social capital is more pronounced
level of economic cooperation. One problem among less educated people, who need to rely
with these measures is that they are contami- more on trust because of their limited under-
nated by other factors. For example, is the level standing of contracting mechanisms.
of trust a New Yorker exhibits in her daily We also examine the mechanism by which
economic behavior the result of good law en- social capital generates the trust needed for fi-
forcement or the product of a high level ofnancial transactions. Is trust simply an equilib-
social capital? We focus on two outcome-based rium outcome of a society in which nonlegal
measures that are free from this criticism: elec-
mechanisms force people to behave coopera-
toral participation and blood donation. There tively (e.g., Coleman, 1990; Spagnolo, 1999) or

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
528 THE AMERICAN ECONOMIC REVIEW JUNE 2004
cruing
is there an inherited component imprinted dur-to people through membership in
certain
ing education? We distinguish between thesecommunities (Pierre Bourdieu, 1986).
Coleman (1990) describes social capital as a
two interpretations by focusing on households
resource of individuals that emerges from so-
that have moved from one province to another.
cial ties. Thus, the source of this capital lies
For these households we can separately identify
the effect of the environment they grew with
upthe
in people a person is related to. But why
are some people willing to make resources
and the environment where they live. Although
most of the effect is due to the level of social available to others without any explicit
capital prevailing in the area where an individ- compensation?
ual lives, a significant fraction (roughly a third) Sociologists identify two main motivations
is due to the level of social capital prevailing in(Alejandro Portes, 1998). First, people may do
the area where he was born. it because of strongly internalized norms (what
Besides the literature on trust and social cap-sociologists call consummatory behavior). They
ital, our work is mostly related to a growing donate to charity, obey traffic rules, and pay
number of studies on the effect of social inter- their debts on time because they feel obli-
action, peer monitoring, and peer pressure on gated to do so. Alternatively, people might be
criminal behavior (Ann Case and Larry Katz, willing to make resources available for instru-
1991), on shirking in the workplace (Andrea mental reasons. In this case, social capital
Ichino and Giovanni Maggi, 2000), on group affects the behavior of individuals because it
lending programs (Timothy Besley and Stephen enhances the level of social punishment of a
Coate, 1995), and on stock market participationsociety.
(Harrison Hong et al., 2004). This literature For most of the paper we do not distinguish
studies the effect of the social structure of small between these two theories. Instead, we focus
groups on economic outcomes. Because we are on a common prediction of both, that high lev-
interested in explaining different patterns of els of social capital generate higher levels of
economic development, we instead look at so- trust toward others. At the end of the paper we
cial characteristics of the whole community try to distinguish whether social capital is
(electoral participation, the incidence of blood purely driven by environmental variables or if
donation) where individuals live and grow up. there is also an inherited component.
We investigate whether the use and availability
of financial instruments is affected by the social B. Social Capital and Financial Development
characteristics of the community.
The paper proceeds as follows. Section I dis-Whether individuals are willing to sign finan-
cusses the notion of social capital and its mea-
cial contracts depends not only on the enforce-
sures. Section II describes the data and the ability of contracts, but also on the extent to
hypotheses that we test. Section III presents the
which they trust the counterpart.
results of the effect of social capital on the Trust
use within a specific group may have am-
and availability of financial contracts. Section
biguous effects on the use of financial contracts.
IV explores situations when social capital is York, diamond traders all belong to a
In New
more important. Section V asks if there is Jewish
an orthodox sect and they do not use con-
inherited component of social capital, imprinted tracts: the within-group trust is sufficient. By
with education. Section VI assesses the relative contrast, trust across groups or generalized trust
importance of the inherited versus environmen- can only benefit the workings of organized mar-
tal component of social capital. Section VII kets and the development of finance.
concludes. Since we focus on social capital at the com-
munity level, we characterize high-social-
I. The Concept of Social Capital capital areas as those with high levels of
generalized trust, which has an unambiguously
A. What Is Social Capital? positive effect on the use and availability of
financial contracts. Thus, we expect financial
In sociology, social capital is broadly de-development to be positively correlated with
fined as the advantages and opportunities ac-our measures of social capital.

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
C. A Simple Framework and higher costs (k) reduce the utility from
absconding.3
To determine our choice of the right proxies Given these assumptions, there is a cost
of social capital and the empirical tests, we
threshold ki = k(XJ, NJ, Oi) below which broker
sketch a simple model of the link between social
i will find it optimal to abscond,
capital, trust, and financial decisions. We con-
struct the model in terms of a household deci-
sion of how much to invest in stock, but we note (2) a { k1 if < ki
that our model can be easily extended to other
(2) a * 0 otherwiseJ'
financial decisions.
Let investor i's demand for stock be repre- Since it can be shown that k is decreasing in
sented by X, N, and Oi, fewer brokers will abscond in areas
with high value of XJ, N', and Oi.
(1) Si= f(ER, (p) We assume for simplicity there are only two
types of agents Oi E (OL, OH), where OH > OL
where Si is the amount of money principal i denotes a type who is less willing to abscond
invests in stock, ER the expected return from with the money. The distribution of broker
the investment, and p,i her individual character- types can differ across areas. Let pJ denote the
istics, such as risk aversion, that affect portfolio frequency of 0L types in the population of
choice. To introduce an element of trust we agents living in area J. In equilibrium, the prob-
assume that the broker will abscond with the ability a broker in area J absconds is given by4
money with some probability.2 If the investor
fears that the broker will abscond with her
(3) 7J = h(p, XJ, NJ).
money, the expected return of her investment
will not simply be R, the expected return on the
stock, but rT X 0 + (1 - 7r) X R where Tr is the This equation represents the probability that
probability the broker will abscond with the an investor in area J will use to compute her
money. expected return. Then, her demand for stock in
How much an investor will trust her broker region J will be
depends on her expectation about the broker's
behavior. To derive this expectation we analyze (4) Si =f((1 - 7Tj) R, (,i)
the broker's decision.
A broker i's utility of absconding can be = I(XJ, NJ, pJ, oi ).
written as Vi = V(a,, XJ, NJ, Oi, ki) where ai E
{0, 1} is the action (abscond or not), XJ the
quality of legal enforcement in area J where the It follows, then, that the demand for stock in
investor is located, NJ the extent of social net- area J will be increasing in the level of legal
works in area J, Oi is the set of social norms ofenforcement (X), in the extent of social network
agent i, and ki an individual-specific fixed cost(N), and in the relative strength of norms (p).5
of absconding drawn from the cumulative dis-
tribution F(k), which is the same across areas. D. How Do We Measure Social Capital?
We assume that higher legal enforcement (X),
broader social networks (N), stronger norms (0), Because we do not observe individual norms,
Oi, and hence pJ, and since it is difficult to
2 The risk of the broker absconding is meant to capture

the various ways trust affects investment decisions. For 3So that Vx < 0; VN < 0; V, < 0, Vk < 0 with V,
instance, in the case of stocks, an investor should not only indicating the partial derivative of V with respect to z.
trust the broker or financial intermediary that buys and 4This comes from vTJ = pJ F(k(XJ, N-, OL)) +
holds the stock for him, but also trust the accounting(1 - pJ)F([k(XJ, NJ, OH)) = h(pJ, XJ, NJ).
numbers the firm reports, the managers running the firm, 5 To derive it is sufficient to notice that sign aS/aX =
etc. The same reasoning applies to any other financialsign aS/aN > 0, since sign 8rr/aX = sign dr/OlN < 0; and
instrument. aS/dp < 0 since dirl/p > 0.

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
observe all the formal and informal social net- (1987), to public order measures (1978,
works, NJ, we follow earlier studies in using 1981).
some outcome-based measures of social capital. Table 1 shows that Italy has a very high
However, to do so we must ensure that these average voter turnout (80 percent). However,
measures are not affected by other environmen- what is relevant for this study is the cross-
tal variables, such as legal enforcement, that are sectional variability, which is substantial. Turn-
uncorrelated with social capital, but which out goes from 62 percent to 92 percent, with
might have a direct impact on our variable ofone-quarter of the observations below 72 per-
interest (e.g., investment in equity), as equation cent and one-quarter above 86 percent. Figure
(4) shows. Therefore, we focus on the choice of 1 shows how this measure of social capital
electoral participation and blood donation. varies within Italy. Social capital is higher in the
Unlike economic cooperation, legal variablesNorth of Italy (north of the Apennines), weaker
and individual characteristics other than norms in the center (from the Apennines to Rome), and
do not enter in the utility of donating blood orvery weak in the South (south of Rome). How-
donating time to vote, Di. By contrast, the utilityever, even within these areas there is some
function might depend on the aggregate (com-variation.
munity) level of donation (or voting), DJ. In Italy 90 percent of the whole blood dona-
Hence, the utility from donating blood can betions and 100 percent of anonymous blood do-
represented as Ui = F(Di, DJ, NJ, Oi). nations are collected by AVIS, the Italian
In equilibrium DJ = f(NJ, pJ). Thus, the localassociation of voluntary blood donors (see the
level of blood donation (or electoral participa- Appendix for more details on AVIS). Since the
tion) is a function of only the two componentscollection procedures are set nationally and ad-
of social capital, networks and norms. If we doministered by AVIS, these data control for pos-
not want to distinguish between these twosible differences in the quality of medical
sources of social capital, DJ is a legitimateinfrastructure. Hence, our second measure of
proxy for the level of social capital in area J. social capital is the number of 16-ounce blood
Hence, we can rewrite (4) as Si = I(XJ, DJ, Spi).bags collected per inhabitant in the province in
We use this specification in our empirical 1995, the only year for which we have complete
analysis. data at the province level.
As Table 1 shows, the average level of do-
II. The Data nation is three bags per hundred people, but
there is high cross-sectional variability. Some
A. Measures of Social Capital provinces have no donations, others go as high
as 11 bags per hundred people.
As noted, our primary measures of social Table 1, Panel B, reports the cross-correlations
between these two measures of social capital.
capital are electoral turnout and blood donation.
We measure both these factors at the province Despite the different nature of these variables,
level.6 their correlation is high (0.64). However, it is
Since in general elections Italian citizens not perfect. Hence, we can gain some insights
are required to vote by the law, we measure by looking at common components.
voter turnout in referenda, where voting is not
mandatory. We measure voter turnout for all B. Measures of Use and Availability of
the referenda that occurred in Italy between Financial Instruments
1946 and 1989. These referenda cover a very
broad set of issues, ranging from the choice Our data on households is drawn from the
between republic and monarchy (1946), di- Survey of Household Income and Wealth
vorce (1974), abortion (1981), from hunting (SHIW). This survey is conducted by the Bank
regulation (1987), to the use of nuclear power of Italy on a representative sample of about
8,000 households. The survey collects detailed
information on Italian household income, con-
6 In our classification Italy is divided in 95 provinces, sumption, and wealth as well as households'
which are similar to U.S. counties. portfolio allocation across financial instruments

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
TABLE 1 SUMMARY STATISTICS AND CORRELATIONS
Panel A: Summary Statistics

Mean Median Standard Minimum Maximum Observations
deviation
Social capital 1 0.8 0.83 0.08 0.62 0.92 32,665

Social capital 1 origin 0.79 0.8 0.09 0.6 0.92 32,184
Participation in referendum on divorce 0.88 0.9 0.07 0.68 0.97 32,583
Social capital 2 0.03 0.02 0.02 0 0.11 32,665
Trust (WVS) 0.33 0.32 0.14 0 0.75 24,674
Use of checks 0.49 0 0.5 0 1 32,665
Percent wealth in cash 0.24 0.06 0.35 0 1 32,332
Percent wealth in stocks 0.03 0 0.12 0 1 32,332
Discouraged or turned down 0.03 0 0.16 0 1 32,665
Loans from friends and family 0.03 0 0.18 0 1 32,665
Judicial inefficiency 3.63 3.5 1.25 1.44 8.32 32,665
Squared judicial inefficiency 14.75 12.26 11.1 2.08 69.28 32,665
North 0.43 0 0.49 0 1 32,665
South 0.36 0 0.48 0 1 32,665
Per capita GDP (in thousands dollars) 14.15 12.86 7 5.21 40.33 32,665
Average years of education 7.54 7.59 0.86 5.75 10.29 32,665
Income (in thousands dollars) 25.3 20.92 18.43 0 428.38 32,665
Wealth (in thousands dollars) 136.38 81.35 233.07 -104.02 9,905.83 32,442
Age 53.1 53 15.17 17 114 32,665
Education 8.2 8 4.67 0 18 32,665
Married 0.74 1 0.44 0 1 32,665
Male 0.78 1 0.42 0 1 32,665
Number of people living in house 3 3 1.37 1 9 32,665
Panel B: Correlations
Social Social Participation Social Trust Judicial Per capita Average

capital 1 capital 1- in divorce capital 2 (WVS) inefficiency GDP years of
origin referendum education
Social capital 1 1
Social capital 1-origin 0.0963 1
(0.0004)
Participation in 0.9711 0.1037 1
divorce referendum (0.0000) (0.0002)
Social capital 2 0.6366 0.0580 0.5864 1
(0.0000) (0.0339) (0.0000)
Trust (WVS) 0.3821 0.1063 0.3876 0.2448 1
(0.0037) (0.0015) (0.0032) (0.0690)
Judicial inefficiency -0.6363 -0.0570 -0.6688 -0.4253 -0.2138 1
(0.0000) (0.0371) (0.0000) (0.0000) (0.1136)
Per capita GDP 0.5466 0.0685 0.5386 0.3686 0.2154 -0.3699 1
(0.0000) (0.0001) (0.0000) (0. (0.0000 (0. ) (0.0012) (0.0000)
Average years of 0.6349 0.1081 0.6635 0.2555 0.3644 -0.5405 0.5416 1
education (0.0000) (0.0001) (0.0000) (0.014) (0.0058) (0.0000) (0.0000)
Notes: The description of the variables is in the Appendix. Panel A contains summary statistics. Pane
among the social capital indicators and other environmental variables. The number in parentheses is the
each correlation coefficient.
and their access to formal and informal credit. One of the unique features of the SHIW is its
For each household, the data also contain infor-ability to distinguish between households that
mation on characteristics of the households' did not want a loan from those households that
did not succeed in obtaining a loan because they
head, such as education, age, place of birth, and
residence. were turned down or did not apply because they

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
d '2
"'rllpalio in h ferenda
in Itaflan provinoes: percentlge rates (number of provnces)
8ormmm (17)
Nteo e (7)
[ 81b 88 (25)
r t72to s1 (21)
3 62t0 72 (22)
FIGURE 1. TRUST ACROSS ITALIAN PROVINCES: PARTICIPATION IN REFERENDA
Note: Darker areas correspond to provinces with a higher participation in referenda.
expected to be turned down. The survey alsosurvey has a rotating panel component, so 9,287
reveals the existence of informal credit (i.e.,of these observations come from reinterviewing
credit extended by friends and family). Thethe same household in a different year. In our
Appendix contains a more detailed description analysis we check the robustness of our results
of the data set, with the actual interview to eliminate these repeated observations. After
questions. excluding a few households that report negative
The SHIW is conducted every two years. consumption and/or income (17 observations),
Since the last four surveys (1989-1995) have our final sample contains 32,665 households if
maintained the same structure, we pool them, we include repeat observations, and 23,330
obtaining a sample of 32,686 observations. The households if we exclude repeat observations.

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
Table 1 reports summary statistics for alleconomic

the development, measured by GDP per
measures of use and availability of financial
capita in the province. This measure is released
instruments: these measures are the dependent by the National Institute of Statistics (ISTAT).
variables in our regressions. The first measure It averages
is 14,000 dollars and exhibits wide
an indicator variable of whether a household variations (standard deviation 7,000 dollars per
uses checks. Half of the households interviewed capita).
do not use checks. The second measure is the The second variable is a measure of the in-
proportion of financial wealth that a householdefficiency of law enforcement, the average
retains in cash. All observations are equallynumber of years it takes to complete a first-
weighted, thus the mean (24 percent) is dis- degree trial in the courts located in the province.
torted by the fact that poorer people retain 100We compute this measure using data released
percent of their financial wealth in cash. Aby the Ministry of Justice on the length of trials.
value-weighted average gives a more reason-As Table 1 indicates, there is wide variation in
able 2.4 percent. This feature highlights thethis measure, ranging from 1.4 to 8.3 years, with
a mean of 3.6 and a standard deviation of 1.25.
importance of controlling for the level of wealth
(and its square to capture possible nonlineari- The third variable is a measure of the average
level of education in the province. Although our
ties) in any regression. The third measure is the
fraction of financial wealth retained in stock. regression controls for the individual level of
The low mean (3 percent) is consistent with the education, the average level of education may
limited role played by the stock market in Italy have important externalities in households' be-
(e.g., Marco Pagano et al., 1998). havior. Therefore, we use the average years of
The next variable pertains to a household's schooling in the province in 1981 (from
ability to access the credit market. "Discour- ISTAT).
aged or turned down" is an indicator variable We know the province where the household
equal to one if a household responds positively currently resides. Accordingly, we merge the
to at least one of the following questions: "Dur- household data set with our measures of social
ing the year did you or a member of the house- capital and attach to each household the mea-
hold think of applying for a loan or a mortgage sures of social capital in the province where it is
to a bank or other financial intermediary, but located. We also know the province where the
then changed your mind on the expectation that household head was born. We use this as a
the application would have been turned down?" proxy for the area in which an individual was
"During the year did you or a member of the raised, and for the level of social capital pre-
household apply for a loan or a mortgage from vailing there. We label this variable social cap-
a bank or other financial intermediary and your ital of origin.
application was totally or partially turned Table 1, Panel B, reports the cross-correlations
down?" Two percent of the sample households between the various measures of social capital
were discouraged from borrowing (i.e., an- and the other environmental variables. As we
swered "yes" to the first question), while 1 expected, social capital measures are positively
percent of the sample households were turned correlated with income per capita and average
down (i.e., answered "yes" to the second education.
question).
"Family loan" is an indicator variable equal D. Theoretical Predictions
to one if a household responds positively to the
question "As of the end of the year did you have All financial contracts could be reduced to a
debts outstanding towards friends or relatives principal who entrusts some money to an agent.
not living with you?" Three percent of the sam- The expected return of this activity depends
ple households received such loans. (among other things) on the probability that the
agent will abscond.
C. Environmental Variables For instance, in accepting a check, the prin-
cipal trusts the agent to have the necessary
funds in the bank. The expected return on the
We augment our household-level data with
several other variables. The first is a measure of check depends on the level of trust in the agent,

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
which is a function of the level of socialparatively

capital.high level of trust (such as friends
Thus, ceteris paribus, households livingand in family).
low-
Third, many authors (Banfield, 1958;
social-capital areas are less likely to use checks.
In a portfolio choice, assets differ not Fukuyama,
only in 1995) emphasize that low levels of
their intrinsic riskiness, but also in thetrust
proba-
toward others are generally associated with
bility of being expropriated, and thushigh in levels
the of trust within subgroups, such as
amount of trust they require. Where social the family.
cap- Banfield's term "amoral familism"
ital (and hence trust) is very low, households
signifies the existence of very high levels of
will invest a larger fraction of their assets trust in
within the family and very low levels out-
the least trust-intensive form of investment, side of it. This phenomenon naturally leads to
i.e., holding cash. Similarly, households willtransactions from the marketplace to
moving
invest a smaller fraction of their assets in the the restricted family circle.
most trust-intensive form of investment, i.e., Given the importance of these three factors,
stock. in low-social-capital areas we expect a higher
Lending is also a trust-intensive activity. Theincidence of loans by friends and family and
lender must trust that the borrower will not run thus a negative correlation between the likeli-
away with the money. Thus, using the same hood of informal loans and the level of social
logic, we expect that the supply of loans to capital.
households is positively affected by the average
level of social capital in the province. Fortu- III. Empirical Results
nately, the data allow us to separate demand and
supply. We have the information on whether the A. Use of Checks
respondent requested a loan and whether he was
turned down or was discouraged from applying. One indicator of the use of financial instru-
Thus, a higher level of social capital should ments is the reliance on checks to settle trans-
decrease the probability that a household is ei- actions. Table 2 reports the probit estimates of
ther denied credit or discouraged from applying. the effect of social capital on the probability that
Our data set also contains information on a household uses checks. We regress the indi-
informal lending, those loans that are made cator
by of use of checks on the level of social
relatives and friends. How do we expect them capital,
to the level of judicial efficiency (linear
vary with the degree of social capital? As for and squared), the GDP per capita, the average
any type of lending, a higher level of trust level of education, several household character-
should lead to more lending. However, in this istics, and three calendar-year dummies. When
case there are three forces pushing in the oppo- we measure social capital at the provincial level
site direction. First, informal lending is a sub-we correct the standard errors for the noninde-
stitute for formal lending when the latterpendenceis of the observations within the same
either unavailable or too expensive. As we note province.
The household characteristics we use are
above, the access to formal lending is jeopar-
dized by lack of social capital. Thus, the household
de- income (linear and squared), house-
mand for loans from friends and family hold wealth (linear and squared), household
increases in areas with low social capital. Sincehead's age (linear and squared), his/her educa-
for these informal loans we cannot separately tion (number of years of schooling), the number
observe the demand and supply, but only their of people in the household, and indicator vari-
existence, it is possible that the demand effect ables for whether the head is married, is a male
dominates and that the likelihood of loans by for the industry in which he/she works, and for
friends and family is higher in areas with the low level of job he/she has.
social capital. Table 2 shows that social capital increases the
Second, there might be a substitution effect probability of using checks. This effect is sta-
on the supply of loans. In low-social-capital tistically significant at the 1-percent level. The
areas, the group with the highest comparative reported coefficients are the effect of a marginal
advantage in undertaking trust-intensive activi- change in the corresponding regressor on the
ties (such as lending) is the group with a com- probability of writing checks. Thus, we can

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
TABLE 2 EFFECT OF SOCIAL CAPITAL ON THE USE OF CHECKS
I II III IV V VI VII
Social capital 1 0.5710*** 0.4265* 0.5552** 1.2584***

(0.1790) (0.2436) (0.2224) (0.3614)
Social capital 1- 0.2078***
origin (0.0481)
Social capital 2 1.8614***
(0.3719)
Trust WVS 0.2196***
(0.0817)
North 0.0941***
(0.0295)
South -0.0078
(0.0397)
Judicial inefficiency -0.0802 -0.0406 -0.0295 -0.0311 -0.0391 -0.0182
(0.0573) (0.0472) (0.0462) (0.0492) (0.0415) (0.0342)
Judicial inefficiency 0.0084 0.0048 0.0042 0.0041 0.0047 0.0037
squared (0.0054) (0.0046) (0.0044) (0.0048) (0.0044) (0.0033)
Per capita GDP 0.0056*** 0.0052*** 0.0045*** 0.0049*** 0.0041*** 0.0019*
(0.0012) (0.0011) (0.0012) (0.0012) (0.0010) (0.0011)
Average years of 0.0570*** 0.0385*** 0.0446*** 0.0656*** 0.0437*** 0.0518***
education (0.0174) (0.0139) (0.0139) (0.0140) (0.0132) (0.0142)
Income 0.0119*** 0.0117*** 0.0116*** 0.0087*** 0.0117*** 0.0117*** 0.0088***
(0.0006) (0.0006) (0.0006) (0.0003) (0.0006) (0.0007) (0.0006)
Income squared -0.0000*** -0.0000*** -0.0000*** -0.0000*** -0.0000*** -0.0000*** -0.0000***
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000)
Wealth 0.2762*** 0.2853*** 0.2915*** 0.1349*** 0.2929*** 0.2864*** 0.1479***
(0.0442) (0.0449) (0.0438) (0.0185) (0.0447) (0.0456) (0.0300)
Wealth squared -0.0363*** -0.0374*** -0.0378*** -0.0160*** -0.0389*** -0.0372*** -0.0187*
(0.0125) (0.0126) (0.0124) (0.0040) (0.0122) (0.0130) (0.0109)
Age 0.0138*** 0.0137*** 0.0136*** 0.0077*** 0.0133*** 0.0135*** 0.0079***
(0.0019) (0.0019) (0.0019) (0.0011) (0.0019) (0.0021) (0.0014)
Age squared -0.0002*** -0.0002*** -0.0002*** -0.0001*** -0.0002*** -0.0002*** -0.0001***
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000)
Education 0.0268*** 0.0268*** 0.0269*** 0.0196*** 0.0267*** 0.0269*** 0.0197***
(0.0017) (0.0017) (0.0017) (0.0007) (0.0017) (0.0020) (0.0012)
Observations 32,442 32,442 32,442 31,961 31,366 32,442 31,366
Pseudo-R2 or R2 0.274 0.276 0.278 0.332 0.278 0.278 0.320
Notes: The dependent variable is an indicator variable that takes a value one if the interviewed household responds positively
to the question "Did you or some other member of the household issue checks in the course of the year to settle transactions?"
For a description of all the other variables see the Appendix. All regressions include as controls family size, dummies for
whether the household head is male, married, for his/her type of job and industry, and calendar-year dummies. Columns III,
V, VI, and VII include as controls four macro-regional dummies (North East, North West, Center, and South). For all columns
except IV and VII the reported coefficients are probit estimates of the effect of a marginal change in the corresponding
regressor on the probability of using a check, computed at the sample mean of the independent variables. The coefficients
reported in column IV are from a linear probability model with fixed province effects. Column VII is estimated by IV, with
social capital 2 as the instrument. The standard errors reported in parentheses are corrected for the potential clustering of the
residual at the provincial level. The symbols ***, **, * mean that the coefficient is statistically different from zero,
respectively, at the 1-, 5-, and 10-percent level.
compute the impact of social capital for anand statistically significant effect on the proba-
individual that moves from the lowest-social- bility of using checks. Since other studies
capital province to the highest-social-capital(Knack and Keefer, 1997; Knack and Zack,
provinces. The probability of using a check2001) show that the level of social capital is
increases by 17 percentage points, about a thirdpositively correlated with economic develop-
of the sample mean. ment, the level of per capita GDP might absorb
The level of per capita GDP has a positive some of the effect of social capital. Nevertheless,

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
we think it is necessary to insert it into

sion the
using a finer partition of the territory into
five are
regression to control for those factors that macroareas: North East, North West, Cen-
associated with financial development, ter,but
South, and Islands, according to Italian Na-
which have nothing to do with social capital.
tional Statistics geographical classification. The
Excluding per capita GDP from the regression
results confirm the previous findings.
(not reported) increases both the size of Allthe
these attempts do not completely elimi-
coefficient of social capital and its statistical
nate the suspicion that some environmental
significance. The average level of education
variables other than social capital might be driv-
also has a positive impact on the probability
ing theofresults. The only way to rule this out
using a check. would be to estimate a model with fixed pro-
vincial effects that can absorb all the factors that
To rule out the possibility that social capital
vary only
is capturing the efficiency of the legal system, in at provincial level. Unfortunately,
all the regressions we control for a measurethese fixed
of effects would also absorb our mea-
sure of social capital.
the quality of the court system. As we expected,
in areas where courts are more inefficient, To solve this problem, we resort to the pres-
households use fewer checks, but this effectence of ismovers in the data. Movers are likely to
not statistically significant. Given the average
be affected not only by the social capital of the
placeare
length of a trial (3.6 years), legal procedures where they live, but also by the social
simply too lengthy to make a difference. All of the place where they grew up. This
capital
other control variables have the expected sign:is present if there is an inherited compo-
effect
age and education increase the probability nent inofsocial capital, or if people form a sub-
using checks, so do income and wealth. jective estimate of trustworthiness based on
In studies by Banfield (1958) and Putnam their past experience. Regardless of the reason,
(1993), the South of Italy is the prototypical
the social capital of origin will have an impact
area deficient in social capital, while theonNorth
the use and availability of financial contract
is richer. Ichino and Maggi (2000) support thisenables us to separate the effect of social
which
view by showing that the degree of shirking byfrom the effect of other environmental
capital
employees of the same bank is significantly variables.7 Therefore, we estimate a linear prob-
higher in the South even after controlling formodel with province fixed effects and the
ability
several characteristics of the employees and
social capital of origin (plus the usual control
those of the individual branches. variables). In this specification (column IV), the
Consistent with these findings, our North- social capital of origin is positive and highly
South indicator variables turn out to be highly statistically significant. This effect cannot be
correlated with social capital. The correlation attributed to omitted variables at the local level.
between the North indicator and our measure of Thus far, we have checked the robustness of
social capital is 60 percent, and there is a neg- our results by using different controls for envi-
ative correlation of 88 percent between the ronmental variables. We now check the robust-
South indicator and social capital. This correla- ness using different definitions of social capital,
tion might generate the suspicion that the effect keeping as geographical controls dummies for
we are capturing is due to some other differ- the five macro regions. Social capital measured
ences between the North and the South of Italy, by blood donation has a positive and statisti-
that just happen to be correlated with our mea- cally significant effect on the probability of
sure of social capital. Controlling for North and using checks (column V). The magnitude of the
South indicators (column II of Table 2), social effect is similar to the one found using electoral
capital still has a positive effect on the proba- participation; the probability of using checks
bility of writing a check, and the effect is sta- increases by 20 percent when moving from the
tistically significant. However, the effect is lowest-social-capital province to the highest-
somewhat smaller: moving from an area with social-capital provinces.
the lowest social capital to an area with the
highest social capital increases the probability
of using checks by 13 percentage points. In 7 We will distinguish among these two different expla-
column III, we reestimate our baseline regres- nations in Section V.

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
In our framework the importance of social errors are slightly bigger. All the results remain
capital on financial development is mediated theby
same.
the level of trust. An obvious way to check our
results is to see if there is a direct relation
between the level of interpersonal trust within a B. Investment in Cash
community and the use and availability of fi-
nancial instruments. We do this in column VI. We use the same specification to estimate the
To measure the level of interpersonal trust, we effects of social capital on portfolio allocation.
rely on the World Values Survey (WVS), whichThe only difference is that we use a two-limit
interviewed varying-sized samples of peopletobit model, since the dependent variable is
across 40 countries, including Italy, in 1990 andconstrained between zero and one. As before,
1999. In each of those surveys, roughly 2,000 we correct the standard errors for possible de-
individuals were asked the question "Generally pendence of observations within the same
speaking, would you say that you trust other province.9
Italians?" Panel A of Table 3 reports the estimated
The WVS is not stratified at the provinceeffects of social capital on the amount of cash
held by a household. Social capital has a nega-
level, thus several provinces are not present and
others are severely underrepresented. To ad- tive and highly statistically significant effect on
dress this problem, we pool the two surveys andthe proportion of wealth a household invests in
cash. A one-standard-deviation increase in so-
we group data at the regional level, by attribut-
ing to each family the average response in thecial capital reduces the amount of cash by 7
region where it is located (the 95 Italian prov-percentage points, a reduction of almost a third
inces are organized in 20 regions). Using thisin the average amount of cash held. Moving
measure of trust we reestimate our basic regres- from the lowest-social-capital province to the
sion.8 The effect of trust has the predicted signhighest-social-capital provinces decreases the
and is statistically significant, though the eco-percentage of wealth held in cash by 27 per-
nomic impact is roughly 30 percent lower than centage points.
the estimates we obtain by using our primary The degree of judicial inefficiency has a non-
measure of social capital. linear effect on the amount of money that
We also take our basic measure of social households retain in cash. This nonlinearity,
capital (electoral participation) and instrument
which is present in most specifications, is con-
it with the blood donation. This method allows
sistent with the role played by courts. At low
us to pool whatever is common to these two levels of inefficiency, small variations can
measures. The estimated coefficient (column VII) have a large impact on portfolio choices, but
doubles, suggesting that the effect is driven by the beyond a certain point, legal enforcement be-
common element in all these three measures. comes inframarginal. A further increase in the
We are also concerned that our sample con- degree of judicial inefficiency has very little
tains some repeated observations. Although the impact.
use of checks changes over time, the residuals The level of per capita GDP has a negative
might be correlated across observations of the effect on the amount retained in cash. This
same individual. Since the cross-sectional cor- effect, which is highly significant, also captures
relation in the residuals is confined to only a some of the relation between social capital and
subset of the observations, and among these to amount retained in cash.
pairs of observations, this correlation is unlikely All other control variables have the expected
to be a problem. But rather than speculate, we sign and most of them are statistically signifi-
reestimate (not reported) all the regressions by cant. Age and education reduce the fraction of
restricting the sample to the first observation offinancial wealth held in cash, as do income and
every household. As we expected, the standard wealth, but at a decreasing rate (the coefficient
8 In this regression we correct standard errors for possi- 9 When we use trust, we correct the standard errors for
ble clustering at the regional level. possible dependence of observations within a region.

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
538 538 ~~~~~THE AMERICAN ECONOMIC REVIEWJUE20 JUNE 2004
TABLE 3-EFFECT OF SOCIAL CAPITAL ON PORTFOLIO SHARES
Panel A: Percentage of Wealth Invested in Cash
-0.8854*** -0.5007***
Social capital -0.5733*** 1 -0.4999*
(0.1582) (0.1824) (0.1755) (0.2764)
Social -0.1961*** 1-
capital
origin (0.0350)
Social capital 2 -0.6112*
(0.3544)
Trust WVS -0.2036***
(0.0720)
North -0.0506***
(0.0146)
South 0.0849**
(0.0371)
Judicial inefficiency 0.1 185*** 0.0860*** 0.0787*** 0.0833*** 0.0864*** 0.0639***
(0.0266) (0.0240) (0.0239) (0.0257) (0.0316) (0.0200)
Judicial inefficiency -0.0134*** -0.0102*** -0.0096*** -0.0095*** -0.0099 -0.0077***
squared (0.0029) (0.0027) (0.0027) (0.0029) (0.0000) (0.0022)
Per capita GDP -0.0006 -0.0003 0.0003 -0.0006 0.0006 0.0001
(0.0007) (0.0008) (0.0009) (0.0011) (0.0000) (0.0009)
Average years of -0.0220* 0.0033 0.0024 0.0037 0.0044 0.0019
education (0.0133) (0.0136) (0.0132) (0.0168) (0.0000) (0.0134)
Income -0.0081*** -0.0079*** -0.0079*** -0.0068*** -0.0078*** -0.0079 -0.0069***
(0.0009) (0.0008) (0.0009) (0.0002) (0.0009) (0.0000) (0.0007)
Income squared 0.0000*** 0.0000*** 0.0000*** 0.0000*** 0.0000*** 0.0000 0.0000***
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000)
Wealth -0.0875*** -0.0926*** -0.0972*** -0.0872*** -0.0989*** -0.0929*** -0.0940***
(0.0292) (0.0291) (0.0296) (0.0135) (0.0294) (0.0345) (0.0257)
Wealth squared 0.0083 0.0091 0.0097 0.0091*** 0.0100 0.0091 0.0099
(0.0100) (0.0099) (0.0100) (0.0029) (0.0098) (0.0110) (0.0088)
Age -0.0067*** -0.0066*** -0.0065*** -0.0053*** -0.0068*** -0.0065 -0.0053***
(0.0013) (0.0012) (0.0012) (0.0008) (0.0013) (0.0000) (0.0011)
Age squared 0.0001*** 0.0001*** 0.000l*** 0.000l*** 0.0001*** 0.0001 0.0001I' *
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000)
Education -0.0103*** -0.0103*** -0.0103*** -0.0091*** -0.0099*** -0.0102 -0.0085***
(0.0015) (0.0015) (0.0015) (0.0005) (0.0015) (0.0000) (0.0011)
Observations 32,332 32,332 32,332 31,851 31,259 32,332 31,259
Pseudo-R 2or R2 0.200 0.204 0.204 0.260 0.204 0.204 0.241
Panel B: Percentage of Wealth Invested in Stocks
Social
1.7380***capital
0.6515 0.9106* 1 0.2303***
(0.3595) (0.5476) (0.5265) (0.0785)
Social capital 1-
0.0473***
origin (0.0129)
Social capital 2 2.5325***
(0.7879)
Trust WVS 0.4061***
(0.1505)
North 0.2267***
(0.0430)
South -0.1890*
(0.1060)
Judicial inefficiency -0.0608 0.0447 0.0707 0.0611 0.0499 0.0069
(0.0959) (0.0774) (0.0757) (0.0820) (0.0693) (0.0045)
Judicial inefficiency 0.0059 -0.0030 -0.0053 -0.0048 -0.0035 -0.0003
squared (0.0105) (0.0100) (0.0097) (0.0107) (0.0000) (0.0004)

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
TABLE 3 Continued.
Panel B: Percentage of Wealth Invested in Stocks-Continued.

Per capita GDP 0.0001 0.0001 -0.0013 -0.0004 -0.0020 -0.0003

(0.0032) (0.0015) (0.0020) (0.0018) (0.0000) (0.0003)
Average years of 0.0280 -0.0506** -0.0462* -0.0234 -0.0469 -0.0019
education (0.0346) (0.0256) (0.0259) (0.0265) (0.0000) (0.0025)
Income 0.0149*** 0.0144*** 0.0142*** 0.0010*** 0.0141*** 0.0143 0.0011***
(0.0013) (0.0013) (0.0012) (0.0001) (0.0012) (0.0000) (0.0002)
Income squared -0.0000*** -0.0000*** -0.0000*** 0.0000*** -0.0000*** -0.0000 0.0000
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000)
Wealth 0.3643*** 0.3767*** 0.3847*** 0.0896*** 0.3775*** 0.3799*** 0.0870***
(0.0547) (0.0527) (0.0533) (0.0050) (0.0530) (0.0630) (0.0125)
Wealth squared -0.0389** -0.0408*** -0.0419*** -0.0091*** -0.0408*** -0.0414 -0.0090***
(0.0152) (0.0149) (0.0149) (0.0011) (0.0146) (0.0000) (0.0031)
Age 0.0162*** 0.0156*** 0.0155*** -0.0001 0.0152*** 0.0154 0.0002
(0.0050) (0.0049) (0.0048) (0.0003) (0.0048) (0.0000) (0.0003)
Age squared -0.0002*** -0.0002*** -0.0002*** -0.0000 -0.0002*** -0.0002 -0.0000
(0.0000 ) (0.0000) (0.0000) (0.0000 (0.0 (0. ) (0.0000 (0. ) (0.0000)
Education 0.0251*** 0.0252*** 0.0253*** 0.0008*** 0.0253*** 0.0250 0.0010***
(0.0020) (0.0020) (0.0020) (0.0002) (0.0020) (0.0000) (0.0003)
Observations 32,332 32,332 32,332 31,851 31,259 32,332 31,259
Pseudo-R2 or R2 0.258 0.267 0.268 0.141 0.269 0.268 0.133
Notes: In Panel A the dependent variable is the proportion of financial

the proportion of financial wealth a household retains in stocks or mutua
see the Appendix. All regressions include as controls family size, dummie
for his/her type of job and industry, and calendar-year dummies. Co
macro-regional dummies (North East, North West, Center, and South
coefficients are tobit estimates. The coefficients in column IV are from a
Column VI is estimated by IV, with social capital 2 as instrument. The sta
for clustering of the residual at the provincial level. The symbols ***, **,
from zero, respectively, at the 1-, 5-, and 10-percent level.
of income squared and iswealth squared

30 percent lower, is
but still highly statistically
positive). significant.
The correlation between low social capital Another possibility is that households retain
and high cash holdings might be due to the their financial wealth in cash to hide it from tax
higher presence of organized crime in areas investigations. Even in this case it would be
with low social capital. To be less visible, indi- surprising that the same people would be will-
viduals involved in criminal activities prefer to ing to reveal this information to the Bank of
retain wealth in cash. However, this objection Italy, which is a government institution. They
ignores the fact that the data come from per- would probably refuse to participate in the sur-
sonal interviews conducted by the Bank of Italy. vey or, if they participate, they would underre-
Thus, it is highly unlikely that an organized port the amount of cash holdings. However, to
crime participant would agree to answer these rule out this possibility, we run the same regres-
questions. However, to rule out this possibility sions by excluding self-employed workers (in-
we control for the level of crime in a separate come underreporting is easier and thus more
regression. We measure crime as the number of widespread among self-employed workers).
violent crimes divided by the population. This The results (not reported) are unchanged.
robustness check also deals with the possible After controlling for the North and South
concern that our measure of judicial inefficiency indicator variables and finer geographical clas-
is an imperfect proxy for law enforcement. The sifications (column II and III), social capital still
estimated effect of social capital (not reported) has a negative and statistically significant effect

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
on the proportion of wealth retained in by cash.

the individual level of risk aversion and it
Thus, the effect of social capital is not perfectlycould be that our social capital measures are in
collinear with the North-South divide. The eco- fact capturing it. Fortunately, the 1995 survey
nomic significance is somewhat lower but still attempts to elicit attitudes towards risk. Each
substantial: moving from the lowest-social-survey participant is offered a hypothetical lot-
capital province to the highest-social-capitaltery and is asked to report the maximum price
provinces decreases the percentage of wealththat he would be willing to pay to participate.
held in cash by 15 percentage points. By using the responses to the question, we are
The social capital of origin has a negative and able to construct an Arrow-Pratt measure of
statistically significant effect on the level of absolute risk aversion for 4,301 households. We
wealth invested in cash (column IV). This resultthus reestimate (not reported) our basic regres-
confirms that the effect of social capital cannotsions for cash and stocks on this subsample,
be attributed to omitted variables at the local including among the regressors the inverse of a
level. measure of relative risk aversion, as implied by
The results are robust to changes in the proxy the solution of a standard portfolio problem
for social capital. Even when measured with
(Robert Merton, 1971). We compute the rela-
blood donation (column V of Table 3, Panel A),
tive risk aversion by multiplying the absolute
risk aversion and the level of the household's
social capital has a negative and statistically
significant effect on the level of cash holdings.
consumption. In both regressions, in spite of the
A one-standard-deviation increase in the level smaller sample, the coefficients of social capital
of blood donation decreases the level of cash preserve the same signs and are still statistically
holdings by 3.7 percentage points, which corre-significant.
sponds to 15 percent of the sample average. Our second concern is that social capital may
Results are similar if we use the WVS trust
be capturing differences in consumers' expo-
measure (column VI), or if we instrument sure
our to uninsurable sources of uncertainty
basic measure of social capital with blood do-
(background risk), which makes them less will-
nation (column VII). ing to buy risky assets. To address this potential
problem we use a section of the survey that
C. Investment in Stock collects data on the subjective probability dis-
tribution of future earning. In the 1995 survey,
Panel B of Table 3 estimates the effect of for half of the sampled households each house-
social capital on the proportion of financial hold member of working age is asked to give a
wealth invested in stock. As predicted, thesubjective
ef- assessment of the probability that
fect is positive and statistically significant.he/she
This will lose his/her job (if employed) or find
result holds when we control for North and one (if unemployed) in the following 12
South (column II), for macro-regional dummies
months. Conditional on being employed, he/she
(column III), when we use blood donation (col-
is then asked to report the minimum and max-
umn V) or trust (column VI) as a measure imumof earnings and the probability that earnings
social capital, or if we instrument our basic
will fall below the midpoint of this range.
measure of social capital with blood donation Following Guiso et al. (2002) we use this
(column VII). information, available for 1,916 households, to
Also, we find that even after controllingcompute
for a measure of expected earnings and
fixed province effects (column IV), the social
their variance. We then reestimate our regres-
sions
capital of origin has a strong positive effect on for cash and stocks now adding these
the proportion of financial wealth invested variables
in scaled by total financial assets (not
stock. The impact is also economically mean- reported). As predicted by theory, earnings vari-
ingful. Moving from the lowest-social-capital ance has a negative effect on the demand for
province to the highest-social-capital provincesstock. More important to our analysis, in all
leads to an increase of 52 percentage points in the sign and significance of the coefficient
cases
the proportion of wealth invested in stock.of social capital is unaffected, indicating that it
We have two concerns with our specification. does not reflect omitted measures of back-
The first is that portfolio allocations are affected
ground risk.

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
TABLE 4--EFFECT OF SOCIAL CAPITAL ON THE AVAILABILITY OF CONSUMER CREDIT
Social capital 1 -0.0588*** -0.0896*** -0.0986*** -0.1957***

(0.0192) (0.0264) (0.0268) (0.0688)
Social capital 1- -0.0365*
origin (0.0189)
Social capital 2 -0.1956***
(0.0690)
Trust WVS -0.0094
(0.0113
North -0.0046
(0.0030)
South -0.0091
(0.0055)
Judicial inefficiency 0.0107*** 0.0097** 0.0086** 0.0086** 0.0089 0.0093*
(0.0040) (0.0042) (0.0043) (0.0042) (0.0056) (0.0052)
Judicial inefficiency -0.0011** -0.0010** -0.0010** -0.0009* -0.0009 -0.0011*
squared (0.0004) (0.0005) (0.0005) (0.0005) (0.0006) (0.0006)
Per capita GDP 0.0005*** 0.0005*** 0.0006*** 0.0005*** 0.0005*** 0.0009***
(0.0002) (0.0001) (0.0001) (0.0001) (0.0002) (0.0002)
Average years of 0.0001 -0.0010 -0.0013 -0.0028 -0.0000 -0.0040
education (0.0019) (0.0021) (0.0021) (0.0024) (0.0027) (0.0031)
Income -0.0001 -0.0000 -0.0000 -0.0002 -0.0000 -0.0001 -0.0002
(0.0001o (. ) (o.oool) (0.00.0) (0.00.0) (0.00.0 ) (00.0001)
Income squared 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000* 0.0000*
(0.0000) (0. 0 (. ) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000)
Wealth -0.0207*** -0.0212*** -0.0218*** -0.0209*** -0.0234*** -0.0214*** -0.0221***
(0.0059) (0.0060) (0.0061) (0.0073) (0.0061) (0.0061) (0.0062)
Wealth squared 0.0018*** 0.0019*** 0.0019*** 0.0020 0.0021*** 0.0019*** 0.0022**
(0.0006) (0.0006) (0.0006) (0.0016) (0.0006) (0.0006) (0.0008)
Age 0.0002 0.0002 0.0002 -0.0013*** 0.0000 0.0002 -0.0014***
(0.0004) (0.0004) (0.0004) (0.0004) (0.0004) (0.0005) (0.0004)
Age squared -0.0000* -0.0000* -0.0000* 0.0000* -0.0000 -0.0000 0.0000**
(0 ) (0.0000 (0.0 (0. ) (0.0000 (0. ) (0.0000) (0.0000) (0.0000) (0.0000) (00000)
Education -0.0000 -0.0000 -0.0000 -0.0000 0.0000 0.0000 0.0000
(0.0003) (0.0003) (0.0003) (0.0003) (0.0003) (0.0003) (0.0004)
Observations 32,442 32,442 32,442 31,961 31,366 32,442 31,366
Pseudo-R2 or R2 0.068 0.069 0.070 0.023 0.071 0.067 0.017
Notes: The dependent variable is an indicator variable taking value one if

to a financial intermediary has been totally or partially turned down for
application would have been turned down; it is zero otherwise. For a descr
All regressions include as controls family size, dummies for whether the
of job and industry, and calendar-year dummies. Columns III, V, VI, and V
(North East, North West, Center, and South). For all columns except IV an
of the effect of a marginal change in the corresponding regressor on the
computed at the sample mean of the independent variables. The coeff
probability model with fixed province effects. Column VII is estimat
The standard errors reported in parentheses are corrected for the pot
level. The symbols ***, **, * mean that the coefficient is statistically dif
10-percent level.
D. Availability of Credit to Consumers turned-dow

ing for a loan.10
Table 4 reports the results of the effect of
social capital on the availability of loans to
social capital on the availability of loansO o We also estima
households. We estimate a probit model of probability of
the effect of social capital on the probability probability of be
of an individual being a discouraged or confirm thos

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
Table 4 shows that social capital hasHowever, a neg- our data set provides us with infor-
ative effect on the probability of not having mation on the presence of informal loans, i.e.,
access to credit. This effect is statistically loanssig-
extended by friends or family members
nificant at the 1-percent level. The reported not living in the same household. As discussed
coefficients in Table 4 show that a one-standard- in subsection D, we expect that informal credit
deviation increase in social capital leads to a might partially substitute for formal credit
0.47-percent decrease in the probability of being wherever the latter is unavailable. Table 5 tests
discouraged or turned down. The probability that this prediction.
an individual is shut off from the credit market We estimate a probit model in which we
decreases by 2 percentage points when he moves regress the likelihood a household has a loan
from the lowest to the highest social capital area. outstanding with friends or relatives on our
To isolate the impact of social capital from measures of social capital and the usual control
other geographical differences, we estimate the variables (income, wealth, their squares, demo-
same regression by controlling for the North and graphic characteristics, etc.). We find that
South indicators (column II) and macro-regional households that come from areas with low so-
indicators (column III). The coefficient of social cial capital are more likely to receive loans from
capital is even larger than the one obtained in friends or relatives. This result is consistent
column I, suggesting that the importance of social with Banfield's (1958) and Fukuyama's (1995)
capital goes beyond geographical differences. claims that low-social-capital societies rely
We test the robustness of our results for other more heavily on naturally high-trust relation-
measures of economic development (not re- ships such as those with friends and family.
ported) and for other measures of judicial inef- This finding is also consistent with individuals
ficiency (not reported).'1 In all cases the effect absorbing these attitudes in the early years of
remains statistically significant. their lives.
Column IV of Table 4 shows that in the linear This effect is statistically significant and
probability model, the social capital of origin economically nonnegligible: moving from the
coefficient is negative and highly statistically lowest- to the highest-social-capital province
significant. decreases the probability that an individual has
Social capital measured by blood donation loans from friends and family by 3 percentage
has a negative and statistically significant effect points, about the same order of magnitude of the
on the probability of being shut down from sample average probability.
credit (column V), after controlling for macro- Once we control for North and South, and for
regional dummies. The magnitude of the effect macro-regional dummies, the effect of social
is similar to the one we obtain when we use capital is virtually unchanged and still highly
electoral participation. Moving from the lowest- significant (column II and III). The same is true
to the highest-social-capital province decreases when we control for other measures of eco-
the probability of not having access to creditnomic by development (not reported), and for other
2 percentage points. The results are similar when measures of judicial inefficiency (not reported).
we use the WVS measure of trust (column VI), These or results are fully supported by the linear
when we instrument our basic measure of social probability model that controls for province
capital with blood donation (column VII). fixed effects (column IV). Social capital mea-
sured by blood donation has a negative and
E. Informal Credit Market statistically significant effect on the probability
of borrowing from friends and relatives (col-
Thus far, we have restricted our analysis to umn V). Moving from areas of the country with
institutional forms of investment and credit. the lowest blood donation to areas with the
highest blood donation decreases the probabil-
ity that an individual borrows from friends or
1 In addition to per capita GDP, as proxies for economic
relatives by 2 percent. The results are similar
development we have used the proportion of households
that own a dishwasher and a personal computer; as an
when we use trust (column VI), or when we
alternative measure of court inefficiency we have used instrument our basic measure of social capital
the
number of pending trials per capita in a province. with blood donation (column VII).

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
TABLE 5 EFFECT OF SOCIAL CAPITAL ON THE INFORMAL CREDIT MARKET
-0.0968*** capital
Social -0.1196*** -0.1157*** 1 -0.1644
(0.0261) (0.0401) (0.0433) (0.1108)
Social capital
-0.0617*** 1-
origin (0.0207)
Social capital 2 -0.1682
(0.1195)
Trust WVS -0.0046
(0.0170)
North 0.0044
(0.0055)
South -0.0046
(0.0070)
Judicial inefficiency -0.0010 0.0018 0.0021 0.0035 0.0021 0.0021
(0.0074) (0.0074) (0.0078) (0.0077) (0.0080) (0.0091)
Judicial inefficiency 0.0001 -0.0002 -0.0002 -0.0003 -0.0001 -0.0002
squared (0.0008) (0.0008) (0.0008) (0.0008) (0.0008) (0.0011)
Per capita GDP 0.0003 0.0003 0.0002 0.0000 0.0000 0.0002
(0.0002) (0.0003) (0.0003) (0.0003) (0.0004) (0.0005)
Average years of 0.0013 -0.0002 -0.0005 -0.0008 0.0018 -0.0012
education (0.0026) (0.0027) (0.0027) (0.0034) (0.0034) (0.0044)
Income -0.0008*** -0.0008*** -0.0008*** -0.0012*** -0.0008*** -0.0008*** -0.0010***
(0.0001) ( 0.0001) (0.0001) (0.0001) (0.0001) (0.0001) (0.0002)
Income squared 0.0000*** 0.0000*** 0.0000*** 0.0000*** 0.0000*** 0.0000*** 0.0000***
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000)
Wealth 0.0072 0.0076 0.0084 0.0181** 0.0083 0.0086 0.0156**
(0.0070) (0.0069) (0.0067) (0.0080) (0.0070) (0.0057) (0.0067)
Wealth squared -0.0012 -0.0012 -0.0014 -0.0036** -0.0014 -0.0014* -0.0033***
(0.0008) (0.0008) (0.0009) (0.0017) (0.0009) (0.0008) (0.0012)
Age 0.0003 0.0003 0.0003 -0.0026*** 0.0001 0.0003 -0.0022***
(0.0006) (0.0006) (0.0006) (0.0005) (0.0006) (0.0007) (0.0005)
Age squared - 0.0000 * * -0.0000 * * -0.0000** 0.0000 * * -0.0000* * -0.0000 * 0.0000 *
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000)
Education -0.0001 -0.0001 -0.0001 0.0001 -0.0001 -0.0000 -0.0001
(0.0002) (0.0002) (0.0002) (0.0003) (0.0002) (0.0002) (0.0003)
Observations 32,442 32,442 32,442 31,961 31,366 32,442 31,366
Pseudo-R2 or R2 0.082 0.082 0.083 0.034 0.082 0.081 0.026
Notes: The dependent variable is an indicator variable that ta

"As of the end of the year did you have debts outstanding to
of all the other variables see the Appendix. All regressions inc
head is male, married, for his/her type of job and industry,
as controls four macro-regional dummies (North East, Nor
the reported coefficients are probit estimates of the effe
probability of being indebted with a relative or friend, co
coefficients reported in column IV are from a linear probabi
by IV, with social capital 2 as instrument. The standard errors
of the residual at the provincial level. The symbols ***, **, *
respectively, at the 1-, 5-, and 10-percent level.
IV. When Does Social Capital Matter

we explore More?
whether the m
varies according to what t
Our results so far have shown a remarkable
and pervasive correlation between the level of A. Social Capital and Legal Enforcement
social capital in an area and the use and avail-
ability of financial contracts. To gain more con- The importance of social capital in enhancin
fidence on the causal nature of this correlation, trust is likely to be larger in areas where la

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
enforcement is not prompt. If it takes more A similar picture emerges if we look at the
than
three years to enforce a contract (as is the caseof social capital on access to credit (Table
effect
6, Panel
in Italy), the willingness to finance a person will B). In areas with weak law enforce-
depend even more crucially on the possibility
ment, ofthe effect of social capital on the proba-
imposing moral sanctions and/or the existence
bility of being turned down after applying for a
of moral norms in a given community. This
loan or discouraged from borrowing has the
suggests that on average, we should expect
expecteda sign, is large (in absolute terms), and
bigger effect of social capital in Italy,iswhere
statistically significant. By contrast, the effect
law enforcement is slow, than in countries
is not significant (and quantitatively very small)
like Sweden or the United States, where law
in areas with better law enforcement (Table
enforcement is more efficient. More im- 6, Panel B, last two columns).
portantly, this reasoning suggests that cross-
Consistently, we find that the effect of social
sectionally, we should expect a highercapital
effect on informal credit is not statistically
of social capital in parts of Italy where law in areas with better law enforce-
significant
enforcement is comparatively worse.ment, Thisbut that it is three times as big and sta-
tistically significant in areas with weak legal
prediction is unique to trust being the channel
enforcement.
through which social capital affects financial
variables. From a policy point of view, this result sug-
To test this predication, in Table 6 we rees-gests that countries that lack social capital
timate our basic specifications, splitting the should compensate for it with better legal en-
sample between provinces with relatively effi- forcement. However, that they should does not
cient judicial systems (judicial inefficiency be- necessarily mean that they do. In fact, countries
low the median of 3.5 years) and provinces with deficient in social capital also have weak legal
relatively inefficient judicial systems (judicial enforcement. For example, in the sample of 28
inefficiency above the median).12 countries in Knack and Keefer (1997), we find a
The first two columns of Table 6, Panel A, correlation of 0.83 between trust and judicial
present the probit estimates of the likelihood of efficiency; this is true also in our sample where
using checks. In areas with better legal enforce- our measure of social capital and judicial inef-
ment, social capital does not have a statisticallyficiency across Italian provinces are negatively
significant impact on the probability of using correlated (correlation coefficient -0.63, Table
checks. By contrast, in areas with weaker legal 1). This correlation might not be a simple coin-
enforcement the effect is three times as large cidence. Putnam (1993) and La Porta et al.
and statistically significant. The difference is (1997a) suggest that the lack of social capital
also statistically significant at the 1-percent may negatively affect the working of institu-
level. tions, thus also the quality of law enforcement.
The remaining columns of Table 6, Panel A, If this were the case, our estimates would
report the tobit estimates of the effect of social grossly underestimate the overall impact of so-
capital on the fraction of financial wealth in- cial capital.
vested in cash and stocks. The effect of social
capital on the fraction of wealth invested in B. Social Capital and Education
stock is three times as large in areas with weak
law enforcement, and this difference is statisti- The extent to which a financial transaction
cally significant at the 1-percent level. Also, for
requires trust should also depend on the level of
education of the individuals involved in the
wealth invested in cash, the impact of social
capital is lower (only two-thirds) where the
transaction. For example, we compare two in-
courts work better, albeit the difference is not
vestors, an educated one, who can read and
statistically significant. understand the fine print of a financial prospec-
tus, and an unsophisticated one, who cannot
understand most of the terms. The inability to
12 We have also tried to insert the product of social fully grasp all the details of the contract in-
volved makes it harder for the unsophisticated
capital and legal enforcement in our basic regressions, with
similar results. investor to discriminate between legitimate in-

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
vestments and frauds. Ceteris paribus, the un- The extreme infrequency of the phenomenon
makes it more subject to confounding effects.
sophisticated investor will require greater trust
to make the same investment. Furthermore, Foranexample, widows may retain the portfolio
investor who does not have the necessary allocation of their deceased spouses, even though
ability or information to make sophisticated they do not have the same level of education. To
financial decisions (e.g., managing his port-see whether this effect plays any role we reesti-
folio) often delegates this function to some-
mate the two regressions restricting the sample to
male household heads. The difference (not re-
body else. Facing an additional delegation
risk, the unsophisticated investor will require
ported) almost doubles, albeit its statistical signif-
more trust to part with his money. Our pre-icance is still below conventional levels.
diction is that the marginal impact of social Overall, our results suggest that social capital
capital on the use of financial contracts is
matters more for less educated people.
higher among uneducated people than among
educated people. V. Does Social Capital Have an Inherited
To test this predication, we split the sample at Component?
the median level of education of the household
head (eight years, corresponding to the end of Is trust simply an equilibrium outcome of a
junior high school).13 society where nonlegal mechanisms force peo-
Table 7 presents the results. The first two ple to behave cooperatively (e.g., Coleman,
columns report the estimates for the two sub- 1990; Spagnolo, 1999), or is there an inherited
samples of the impact of social capital on the component, imprinted with education? Our fixed-
probability of using a check. The impact of effects results already suggest the existence of an
social capital among less educated people is inherited component. Given the importance of this
eight times as big as the impact of social capital aspect we explore it in greater depth.
among highly educated people. This difference One possible objection to our fixed-effect
is statistically significant at the 1-percent level. estimates is that movers differ from nonmovers
In fact, social capital has no statistically signif- in many dimensions. We are particularly con-
icant impact among highly educated people. cerned that the social capital of origin might act
As we can see in the third and fourth col- as a proxy for some other (unobservable) indi-
umns, the impact of social capital on the pro- vidual characteristics that affect an individual's
portion of wealth invested in cash is three times level of trust. After all, movers are not randomly
distributed. As Table 8, Panel A, shows, 25
larger for low-educated households than for
highly educated households. The difference is percent of the movers move from the South to
statistically significant at the 1-percent level.the North, but only 4 percent move in the op-
Also, the proportion of wealth invested posite
in direction. Since the South of Italy is
stock (last two columns) is more sensitivepoorer,
to migration from South to North might be
social capital among less educated people. less "voluntary," than from North to South. In
However, the difference is quantitatively smallother terms, if a person is "starving," she might
decide to move even if she is very risk averse.
(only 20 percent) and is not statistically signif-
icant. This result is surprising, because we ex-By contrast, in less desperate conditions, only
the least risk-averse people will choose to
pected the effect to be stronger for equity
investments, which require much more knowl- move. If this story holds, conditional on being a
edge to be analyzed. This weak result might mover
be one is more risk averse if she moves
due to the small number of less educated fam- from the South than if she moves from the
ilies who own stock (3.6 percent versus 15North. Since the South tends to have lower
social capital, movers with a lower social cap-
percent of the well-educated families and a pop-
ulation average of 7 percent). ital of origin might be more risk averse. If we do
not fully control for individual risk aversion,
this correlation might explain our results on
13 Since for many years this was the mandatory level of
schooling, there are many people at that level, which we
portfolio holdings and possibly on use of
checks.
include in the low-education group. Hence, the higher num-
ber of observations in this subsample. Unfortunately, we do not have enough

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
TABLE 6-SOCIAL CAPITAL AND LAW ENFORCEMENT
Panel A
I II III IV V VI
Use of checks
Efficient Inefficient Efficient Inefficient Efficient Inefficient
Social capital 1 0.4022 0.8537*** -0.7593*** 1l.0525*** 0.8600** 2.8714***

(0.2555) (0.1699) (0.1783) (0.3300) (0.4011) (0.6548)
Judicial inefficiency 0.5341 -0.0552 -0.3363** -0.0304 1.6656*** -0.2872
(0.4055) (0.0870) (0.1713) (0.0641) (0.5861) (0.2335)
Judicial inefficiency -0.1020 0.0070 0.0729** -0.0006 -0.3189*** 0.0288
squared (0.0724) (0.0074) (0.0338) (0.0056) (0.1063) (0.0208)
Per capita GDP 2.9108*** 2.1576*** -0.1631 -0.3703 1.8944 -5.5639***
(0.9678) (0.6529) (0.5226) (0.7208) (1.8612) (1.9456)
Average years of education 0.0513*** 0.0172 -0.0273** 0.0040 -0.0193 0.0026
(0.0187) (0.0197) (0.0128) (0.0319) (0.0266) (0.0514)
Percentage of households 0.4434 0.5868* -0.3234 -0.0590 0.9311 0.0018
with mobile phone (0.6209) (0.3153) (0.4877) (0.3405) (1.4442) (0.7650)
Income 0.0109*** 0.0133*** -0.0060*** -0.0134*** 0.0129*** 0.0241***
(0.0006) (0.0011) (0.0006) (0.0017) (0.0012) (0.0021)
Income squared -0.0000*** -0.0001*** 0.0000*** 0.0001*** -0.0000*** -0.0001***
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000)
Wealth 0.2550*** 0.3960*** -0.0443* -0.2609*** 0.3810*** 0.4736***
(0.0568) (0.0524) (0.0244) (0.0545) (0.0510) (0.1140)
Wealth squared -0.0235*** -0.0794*** 0.0027 0.0432*** -0.0356*** -0.0684***
(0.0081) (0.0149) (0.0075) (0.0144) (0.0132) (0.0232)
Age 0.0137*** 0.0107*** -0.0063*** -0.0053** 0.0153** 0.0124
(0.0028) (0.0023) (0.0015) (0.0022) (0.0060) (0.0083)
Age squared -0.0002*** -0.0001*** 0.0001*** 0.0001*** -0.0002*** -0.0002*
(0.0000) (0.0000) (0.0000) (0.0000) (0.0001) (0.0001)
Education 0.0236*** 0.0273*** -0.0068*** -0.0126*** 0.0259*** 0.0209***
(0.0023) (0.0019) (0.0016) (0.0022) (0.0019) (0.0050)
Observations 17,198 15,198 17,144 15,142 17,144 15,142
Pseudo-R2 0.2424 0.2839 0.2641 0.1558 0.2424 0.2777
Panel B
I II III IV
Discouraged or turned down Loans from family and friends
Efficient Inefficient Efficient Inefficient
Social capital 1 -0.0010 -0.1338*** -0.0374 0.1543***

(0.0276) (0.0243) (0.0337) (0.0582)
Judicial inefficiency -0.0491** 0.0037 0.0689 (0.041452
(0.0199) (0.0144) (0.0565) (0.0199)
Judicial inefficiency 0.01 10*** (0.0005 (0.0113 0.0035*
squared (0.0037) (0.0013) (0.0104) (0.0018)
Per capita GDP 0.0807 0.4497*** 0.0381 0. 1292
(0.1296) (0.1237) (0.1454) (0.2587)
Average years of education -0.0013 0.0056** (0.0005 0.0054
(0.0016) (0.0026) (0.0022) (0.0060)
Percentage of households -0.0428 0.0992 0-0818 (0.0640
with mobile phone (0.0588) (0.0619) (0.1097) (0.0797)
Income 0.0000 -0.0002 -0.0005* * -0.0013***
(0.0001) (0.0001) (0.0001) (0.0002)
Income squared 0.0000 0.0000 0.0000*** 0.00002*
(0.0000) (0.0000) (0.0000) (0.0000)
Wealth -0.0189*** 0.0267** 0.0023 0.01 83
(0.0067) (0.0114) (0.0079) (0.0127)
Wealth squared 0.00160* 0.0035** -0.0006 -0.0032
(0.0006) (0.0017) (0.0008) (0.0022)

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
TABLE 6-Continued.
Panel B-Continued.
I II III IV
Discouraged or turned down Loans from family and friends
Efficient Inefficient Efficient Inefficient
Age 0.0000
0.0005 -0.0002 0.0009
(0.0006) (0.0006) (0.0009) (0.0007)
Age squared -0.0000 -0.0000* -0.0000 -0.0000***
(0.0000) (0.0000) (0.0000) (0.0000)
Education -0.0001 0.0001 0.0000 -0.0001
(0.0005) (0.0003) (0.0003) (0.0004)
Observations 17,198 15,198 17,198 15,198
Pseudo-R2 0.0671 0.0757 0.0795 0.0922
Notes: This table reestimates the basic regressions

systems (judicial inefficiency below the median)
ciency above the median). Judicial inefficiency is
in the local courts. The left-hand-side variables i
all the other variables see the Appendix. All regre
head is male, married, for his/her type of job and
B report probit, while columns III, IV, V, and VI of
are estimates of the effect of a marginal change i
denied credit (the sum of the probability of being
friends and family, computed at the sample mean
are corrected for the potential clustering of th
coefficient is statistically different from zero, res
information to undertake a our

geneity is unlikely to explain full ana
results and
there seems
what causes people totomove.
be an inherited component
We only in
that these people were
social capital. born in a dif
province than the one they are living i
We have also no way of knowing ho
VI. Whyin
they have been living Does Social
a Capital Matter?
province dif
than the one of birth, or what their ch
istics were before Having
they addressed this problem, we try to
moved. Never
we can make some inferences on the cause of disentangle the relative magnitude of the "envi-
their move on the basis of where they areronmental" component of social capital versus
the "inherited" component. To do so, we create
coming from and where they are going. If the
unobserved characteristics behind the deci-two separate measures of social capital. One is
sion to move drive our results, we should our measure of social capital for the province of
observe very different estimates in the two birth (referenda turnout in the province of
groups of movers. birth), the other is the measure of social capital
For these reasons, in Table 8, Panel B, not for the province of residence (referenda turnout
only do we insert a dummy for movers, but also in the province of residence). To allow for pos-
we decompose the effect of social capital of sible differences between movers and nonmov-
origin on the basis of where an individual is ers, we introduce a separate measure of social
moving from. The results show that on average, capital for the households that did not move.
movers do not behave differently from nonmov- This measure is referenda turnout for the prov-
ers. More importantly, the effect of social cap- ince of residence, which by construction coin-
ital of origin for the movers from the South is cides with the province of birth.
no different from that of movers from the In Table 9 we reestimate all the households'
North. Hence, unobserved individual hetero- regressions by introducing these three variables.

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
TABLE 7-SOCIAL CAPITAL AND EDUCATION
I II III IV V VI
Use of checks
Low High Low High Low High
education education education education education education
Social capital 1 0.7131*** 0.0135 -1.0348*** -0.3442*** 1.8479*** 1.5451***

(0.1678) (0.1512) (0.2088) (0.0716) (0.4988) (0.3412)
Judicial inefficiency -0.0830 -0.0494 0.1456*** 0.0518*** 0.1385 -0.1401
(0.0507) (0.0546) (0.0341) (0.0161) (0.1202) (0.0998)
Judicial inefficiency 0.0091* 0.0048 -0.0166*** -0.0056*** -0.0190 0.0147
squared (0.0048) (0.0053) (0.0038) (0.0017) (0.0144) (0.0109)
Per capita GDP 3.4802*** 0.8719 -0.3212 -0.2020 -2.1051 0.9984
(1.0105) (0.5968) (0.5405) (0.2723) (2.3360) (1.5415)
Average years of 0.0525*** 0.0348** -0.0236 -0.0105 0.0474 0.0097
education (0.0182) (0.0147) (0.0160) (0.0077) (0.0422) (0.0287)
Percentage of households 0.0537 0.4062 0.0539 -0.0299 0.2372 -0.9531
with mobile phone (0.4075) (0.3222) (0.4167) (0.1808) (1.2070) (1.0645)
Income 0.0132*** 0.0073*** -0.0132*** -0.0041*** 0.0273*** 0.0116***
(0.0010) (0.0007) (0.0018) (0.0005) (0.0021) (0.0011)
Income squared -0.0001*** -0.0000*** 0.0001*** 0.0000*** -0.0001*** -0.0000***
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000)
Wealth 0.4680*** 0.0957** -0.3275*** -0.0200 0.7698*** 0.2395***
(0.0612) (0.0391) (0.0595) (0.0149) (0.1222) (0.0584)
Wealth squared -0.1021*** -0.0150** 0.0792*** 0.0021 -0.1452** -0.0243**
(0.0290) (0.0059) (0.0293) (0.0038) (0.0564) (0.0109)
Age 0.0105*** 0.0128*** -0.0050*** -0.0050*** 0.0129* 0.0152**
(0.0017) (0.0025) (0.0017) (0.0013) (0.0073) (0.0062)
Age squared -0.0001*** -0.0001*** 0.0000*** 0.0000*** -0.0001** -0.0002**
(0.0000) (0.0000) (0.0000) (0.0000) (0.0001) (0.0001)
Education 0.0388*** 0.0055* -0.0188*** -0.0011 0.0480*** 0.0068*
(0.0022) (0.0031) (0.0023) (0.0011) (0.0086) (0.0040)
Observations 22,433 9,963 22,353 9,933 22,353 9,933
Pseudo-R2 0.2494 0.1148 0.1725 0.9937 0.1827 0.2478
Notes: This table reestimates the basic regressions for the use of financial in
level of education of the household's head. A household is defined low educa
education. Correspondingly, a household is defined as highly educated if the
left-hand-side variables are as defined in Tables 2 and 3. For a descriptio
regressions include as controls family size, dummies for whether the househ
and industry, and calendar-year dummies. The first two columns' reported co
change in the corresponding regressor on the probability of using checks,
variable. The remaining columns report tobit estimates. The standard error
potential clustering of the residual at the provincial level. The symbols ***
different from zero, respectively, at the 1-, 5-, and 10-percent level.
The pattern of all the results may

is similar.
hold inIn general,
all the since the o
specifications, the social capital of origin
social capital forhasmovers is almo
the same sign as the social capital of residence.
the effect of social capital for n
In four out of seven cases it is statistically
regressions.
significant at conventional levels.
The With only one in which th
only exception
exception, the social capital ofof origin
residence is al-more than tha
matters
ways more important, representing between
the likelihood of 63
receiving a loan
percent and 98 percent of the andoverall effect
friends. of
This result is not su
social capital (i.e., the sum of
thethe effectof
network offriends
the and famil
social capital of origin and the
the social capitalan
area where ofindividual gr
residence). We think that wherethis decomposition
she currently lives.

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
TABLE 8-MOVERS
Panel A
Area of residence
Area of origin North Center South Total

North 2,428 446 327 3,201
27.9 5.12 3.76 36.78
Center 852 848 97 1,797
9.79 9.74 1.11 20.65
South 2,093 699 914 3,706
24.05 8.03 10.5 42.58
Total 5,373 1,993 1,338 8,704
61.73 22.9 15.37 100
Panel B
I II III IV V
Checks Cash Stocks Discouraged Loans from

or turned family and
down friends
Social capital 0.1797** -0.1750*** 0.0273 -0.0321 -0.0606

--origin (0.0863) (0.0629) (0.0233) (0.0340) (0.0371)
Social capital- -0.0085 0.0059 -0.0040 -0.0006 -0.0026
origin*South (0.0223) (0.0162) (0.0060) (0.0088) (0.0096)
Movers 0.0001 0.0007 -0.0026 0.0025 0.0038
(0.0063) (0.0046) (0.0017) (0.0025) (0.0027)
Income 0.0087*** -0.0068*** 0.0010*** -0.0002 -0.0012***
(0.0003) (0.0002) (0.0001) (0.0001) (0.0001)
Income squared -0.0000*** 0.0000*** 0.0000*** 0.0000 0.0000***
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000)
Wealth 0.1348*** -0.0870*** 0.0893*** -0.0208*** 0.0183**
(0.0185) (0.0135) (0.0050) (0.0073) (0.0080)
Wealth squared -0.0160*** 0.0090*** -0.0091*** 0.0020 -0.0036**
(0.0040) (0.0029) (0.0011) (0.0016) (0.0017)
Age 0.0077*** -0.0053*** -0.0001 -0.0013*** -0.0026***
(0.0011) (0.0008) (0.0003) (0.0004) (0.0005)
Age squared -0.0001*** 0.0001*** -0.0000 0.0000* 0.0000**
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000)
Education 0.0196*** -0.0091*** 0.0008*** -0.0000 0.0001
(0.0007) (0.0005) (0.0002) (0.0003) (0.0003)
Observations 31,961 31,851 31,851 31,961 31,961
R2 0.332 0.260 0.141 0.023 0.034
Notes: In this table we analyze the behavior of the m

provinces, Panel A shows the transition matrix betw
B reports coefficients from a linear probability
left-hand-side variables are as defined in Tables 2, 3,
variables see the Appendix. All regressions includ
whether the household head is male, married, fo
calendar-year dummies. The standard errors repor
potential clustering of the residual at the provin
that the coefficient is statistically different fro
10-percent level.
cannotthat
In this analysis we assume exclude that
people peop
move
for reasons that have areas
nothingwhere the
to do commu
with the
level of social capital in the area.
capital However,
is similar we
to their

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
TABLE 9-WHY DOES SOCIAL CAPITAL MATTER?
Panel A
II III IV V
Checks Cash Stocks Discouraged or Loans from
turned down family and frie
Social capital 1 for 0.4418*** -0.7603*** 0.0902*** -0.0785*** -0.1327***
nonmovers (0.1302) (0.1268) (0.0259) (0.0240) (0.0340)
Social capital 1 of origin 0.1778*** -0.1912*** 0.0379*** -0.0273 -0.0755**
for movers (0.0527) (0.0706) (0.0122) (0.0183) (0.0315)
Social capital 1 of 0.2857** -0.5784*** 0.0492* -0.0485 -0.0517
residence for movers (0.1313) (0.1332) (0.0260) (0.0309) (0.0434)
Judicial inefficiency -0.0616 0.0977*** -0.0022 0.0098** -0.0019
(0.0430) (0.0235) (0.0068) (0.0046) (0.0093)
Judicial inefficiency 0.0065 -0.0109*** 0.0003 -0.0010* 0.0002
squared (0.0041) (0.0026) (0.0006) (0.0005) (0.0011)
Per capita GDP 2.2724*** -0.4388 0.0940 0.3532** 0.1944
(0.5079) (0.3715) (0.2192) (0.1452) (0.1675)
Average years of education 0.0424*** -0.0195 -0.0010 -0.0009 0.0011
(0.0140) (0.0123) (0.0025) (0.0020) (0.0034)
Percentage of households 0.1040 0.0058 -0.0357 0.0915* -0.0484
with mohile phone (0.2944) (0.2735) (0.0641) (0.0491) (0.0813)
Income 0.0091*** -0.0072*** 0.00l1*** 0.0002* -0.0012***
(0.0006) (0.0007) (0.0002) (0.0001) (0.0002)
Income squared -0.0000*** 0.0000*** 0.0000 0.0000* 0.0000***
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000)
Wealth 0.1361*** -0.0792*** 0.0868*** 0.0181*** 0.0195***
(0.0309) (0.0255) (0.0126) (0.0061) (0.0065)
Wealth squared -0.0165 0.0074 -0.0088*** 0.0016* -0.0038***
(0.0110) (0.0089) (0.0030) (0.0008) (0.0012)
Age 0.0078*** -0.0054*** -0.0001 -0.0014*** -0.0026***
(0.0013) (0.0011) (0.0003) (0.0004) (0.0005)
Age squared -0.0001*** 0.0001*** -0.0000 0.0000** 0.0000**
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000)
Education 0.0196*** -0.0087*** 0.0008*** 0.0000 0.0000
(0.0011) (0.0011) (0.0003) (0.0003) (0.0003)
Observations 31,961 31,851 31,851 31,961 31,961
R 2 0.319 0.238 0.130 0.016 0.025
Panel B
I II III IV V
turned down family and frie
Social capital 1 for 0.4833*** -0.7431*** 0.1246*** -0.0964*** -0. 1400** *
nonmovers (0.1526) (0.1199) (0.0341) (0.0326) (0.0412)
0.2634*
Social capital 1 of origin -0.1694 0.0938** -0.1292 -0.1830***
for movers (0.1370) (0.1742) (0.0395) (0.0832) (0.0625)
Social capital 1 of 0.2570 -0.5855*** 0.0317 0.0318 0.0426
residence for movers (0.1749) (0.1910) (0.0486) (0.0776) (0.0615)
Judicial inefficiency -0.0725 0.0904*** -0.0016 0.0132** -0.0024
(0.0476) (0.0233) (0.0083) (0.0060) (0.0102)
Judicial inefficiency 0.0076* -0.0104*** 0.0003 -0.0013* 0.0001
squared (0.0045) (0.0026) (0.0008) (0.0007) (0.0012)
Per capita GDP 2.3066*** -0.6369 0.0650 0.4073*** 0.1142
(0.5661) (0.4739) (0.2770) (0.1452) (0.1718)
Average years of education 0.0462*** -0.0213* -0.0020 -0.0018 0.0008
(0.0158) (0.0123) (0.0031) (0.0024) (0.0036)
Percentage of households 0.1110 -0.0418 -0.0574 0.1031 -0.0757
with mohile phone (0.3116) (0.2669) (0.0757) (0.0662) (0.0891)

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
TABLE 9 Continued.
Panel B Continued.
I II III IV V
turned down family and frien
Income 0.0083*** -0.0064*** 0.0010*** -0.0001 -0.0011***
(0.0007) (0.0007) (0.0002) (0.0001) (0.0002)
Income squared -0.0000*** 0.0000*** 0.0000 0.0000 0.0000***
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000)
Wealth 0.1426*** -0.0647*** 0.0929*** -0.0169** 0.0285***
(0.0334) (0.0231) (0.0134) (0.0071) (0.0083)
Wealth squared -0.0156 0.0047 -0.0081*** 0.0014 -0.0042***
(0.0117) (0.0079) (0.0027) (0.0009) (0.0013)
Age 0.0114*** -0.0035** 0.0002 -0.0018*** -0.0026***
(0.0017) (0.0015) (0.0004) (0.0005) (0.0006)
Age squared -0.0001*** 0.0000*** -0.0000 0.0000** 0.0000
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000)
Education 0.0204*** -0.0078*** 0.0009*** 0.0001 0.0002
(0.0013) (0.0013) (0.0003) (0.0004) (0.0003)
Observations 23,223 23,141 23,141 23,223 23,223
R2 0.291 0.233 0.138 0.015 0.023
Notes: In this table we modify the way in w

that moved across provinces, we differentia
province of residence. Then, we have the
defined in Tables 2, 3, 4, and 5. For a des
controls family size, dummies for whether
calendar-year dummies. In Panel A all t
regressions are estimated by instrumenta
standard errors reported in parentheses ar
The symbols ***, **, * mean that the coe
10-percent level.
our results will underestimate the effect of so- rule out that discrimination might play a role,
cial capital of origin relative to that of we can rule out that discrimination is the only
residence. source of this effect. In fact, it would be hard to
A bias will arise only if people have a pref- argue that individuals born in areas with low
erence for living with others who share the same social capital hold more cash and less stock as a
set of values. Under this hypothesis, people also result of discrimination, as columns I and III of
tend to choose a spouse with a similar set of Table 9 indicate.
values. Hence, we can use the social capital of Furthermore, if discrimination plays a very
origin of the spouse as an instrument for the big role in the relation between social capital
unobservable component of values of each head and the use of financial contracts, the overall
of household. The instrumental variable esti-
effect of social capital for movers should be
mates are reported in Table 9, Panel B. As much bigger than the effect of social capital for
expected, the IV estimates of the social capital nonmovers who do not face discrimination.
of origin tend to be higher in absolute value than This conjecture is not confirmed by our results.
are those of the social capital of residence,The sum of the effects of the two social capital
albeit noisier. measures for movers is almost identical to the
One possible objection to our interpretation total effect for nonmovers.
that the social capital of origin affects the use An alternative interpretation that would ex-
and availability of financial contracts is that the plain some of our results is that movers are
estimated coefficients might be capturing the unable to assess immediately the extent of local
effects of discrimination. Although we cannot networks and norms in their new area of

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
TABLE 10-THE EFFECT OF TRUST ON FINANCIAL DEVELOPMENT AROUND THE WORLD
External equity Number of Number of Debt over Percent of companies

over GNP domestic firms IPOs GNP publicly held
Dependent variable over population over population
Log per capita GNP 0.026 -1.486 0.049 0.994** 0.144***
(0.475) (1.856) (0.174) (0.040) (0.041)
Trust 0.011* 0.470** 0.054*** 0.003 0.012**
(0.005) (0.204) (0.019) (0.004) (0.005)
Rule of law -0.055 1.701 -0.201 0.029 -0.020
(0.035) (1.365) (0.136) (0.031) (0.038)
Constant 0.115 7.523 -1.352 -0.958** - 1.751***
(0.534) (20.860) (2.005) (0.455) (0.590)
R2 0.14 0.47 0.4 0.44 0.48
Observations 30 30 27 28 30
Notes: The dependent variables are

measure is the fraction of the capi
divided by GNP. The second measu
is the number of initial public offe
total debt outstanding divided by G
percent as a threshold. The data on
with data from Eurobarometer. In
per capita GNP is from La Porta et
assessment of the law and order tr
Porta et al. (1997a). All the coeffi
parentheses. The symbols ***, **, *
and 10-percent level.
residence. Hence, they

ual characteristics imprinted with education, m
networks andwhich persist when peoplein
norms ar move. This
the result
born as is consistent
initial prior with Ichino and
andMaggi (2000),
upd
more. This hypothesis
who find that the shirking behavioris con
of south-
our findings, but
ern employees cannot
persisted after they moved to ex
uals coming from
the North. a low-s
are denied credit more fr
denial of credit VII. Conclusions
does not de
cant expectations, but on
expectations about the
Our findings show that trust
social capital plays an
applicant. To ascertain
important wh
role in the degree of financial devel-
lation exists even
opment across exclud
different parts of Italy. Social
borrowers, capital
we seems reestimate
to matter the most when education
probability levels
of are low and law enforcement isdenie
being weak.
the households
This is preciselywho were
the situation in many develop-
find that it is still
ing countries. true
The obvious question is how that
gen-
origin positively affects
eralizable these results th
are. Is this just a feature
ing denied credit. This
of a country with inefficient resu
legal enforcement?
only do moversIs it an effectexpect othe
we can find only in a microeco-
according to nomic
their initial
analysis that does not have any aggregate pr
people expect them
consequences? to be
that prior. Thus,
We cannot fully rule a slow
out the first possibility.ad
tations alone is
In fact, not
our analysis suffic
of the interaction between
results. trust and legal enforcement suggests that trust is
To fully explain these
much less important (sometimes re
not important
resort to some intrinsic differences in individ- at all) in areas where the court system is more

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
efficient or where people are more educated. sure

As of trust and several indicators of financial
development used by La Porta et al. (1997b).
a result, we could certainly question the impor-
These indicators are the ratio of stock market
tance of social capital in highly developed coun-
tries, where there is good legal enforcementcapitalization
and to GDP, the number of listed
a high level of education. However, most ofcompanies
the per million of population, the num-
ber of IPOs per million of population, and the
world does not fit this description. Hence, social
diffusion of corporate ownership. Although this
capital is likely to be very important in explain-
ing the success (or lack thereof) of developing
finding is far from a definitive proof, it suggests
countries. that our results may extend beyond a single
Instead, we try to answer the second question.
country.
Unfortunately, we do not have cross-country If they do, then the question of how to ad-
measures of social capital to replicate our dress
re-deficiencies in social capital becomes of
gressions. However, Knack and Zack (2001) great policy relevance. Our analysis suggests
that better law enforcement and greater edu-
report an aggregate measure of trust by country,
which they derive from the World Values cationSur- can possibly eliminate the negative
vey. As Table 10 indicates, after controllingeffects
for of lack of social capital. Only future
the degree of law enforcement and the level research,
of however, will be able to tell how to
remove the ultimate causes of social capital
GNP per capita, we find a positive and statisti-
deficiencies.
cally significant correlation between this mea-
APPENDIX: DATA SOURCES AND VARIABLES DESCRIPTION
Our main data source is the Bank of Italy Survey of Household Income and Wealth (SHIW),
which collects detailed data on demographics, household consumption, income, and balance
sheets. We use four waves (1989, 1991, 1993, 1995) because sample size and design, sampling
methodology, and questionnaire content are unchanged. Each survey covers more than 8,000
households for a total of 32,648 household-year observations. Each SHIW surveys a represen
tative sample of the Italian resident population. Sampling is in two stages, first municipalitie
and then households. Households are randomly selected from registry office records. House-
holds are defined as groups of individuals related by blood, marriage, or adoption, and sharing
the same dwelling. The head of the household is conventionally identified with the husband, if
present, otherwise with the person responsible for managing the household's resources. Andrea
Brandolini and Luigi Cannari (1994) present a detailed discussion of sample design, attrition,
and other measurement issues, and comparisons of the SHIW variables with the correspondin
aggregates. Starting in 1989, each SHIW has reinterviewed some households from the previou
surveys. The panel component has increased over time. The SHIW reinterviewed 15 percent of
the previous survey sample in 1989, 27 percent in 1991, 43 percent in 1993, and 45 percent in
1995. In the panel component, the sampling procedure is also determined in two stages:
selection of municipalities (among those sampled in the previous survey), and then selection of
households reinterviewed. This implies that there is a fixed component in the panel (for instance,
households interviewed five times between 1987 to 1995, or four times from 1991 to 1995) and
a new component in every survey (for instance, households reinterviewed only in 1989). The
SHIW has been supplemented with geographical data on social capital, judicial inefficiency, an
economic development.

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
TABLE Al-VARIABLE DESCRIPTION AND DATA SOURCES
Variable Description Source

Social capital 1 level for all the referenda before our
Voter turnout at the province Ministry of Interior
household data start (1989). These include data referenda on the period
between 1946 and 1987. For each province turnout data were averaged
across time.
Social capital 1 1 origin
The measure of social capital in the province of birth of the household Ministry of Interior
head.
Participation in level for the divorce referendum (June 1978).
Voter turnout at the province Ministry of Interior
referendum on
divorce
Social capital 2 Number of blood bags (each bag contains 16 ounces of blood) per million AVIS
inhabitants in the province collected by AVIS, the Italian association of
blood donors, in 1995 among its members. The association, which is
completely private and nonprofit, was founded in the early 1920's and is
present in all Italian regions and 91 provinces (out of 95) with 2,796 city
branches. It groups about 875,000 donors and is the largest blood
donors' association not only in Italy where it collects over 90 percent of
the whole blood donation, but also in the world. Its members who work
for it voluntarily run the association. Blood donations are completely
anonymous. All the blood collected is handed over freely to the public
hospitals. Beneficiaries remain anonymous both to the donors and to the
association. The four provinces where there is no AVIS local branch
have presumably very low or zero blood donations. In the reported
regressions we exclude the four provinces that have no AVIS branch.
However, our results are not affected by this exclusion.
Trust (WVS) An index of the level of trust based on the WVS for Italy run among 2,000 World Values
individuals in years 1990 and 1999. The question asked to the Survey
respondent was: "Using the responses on this card, could you tell me
how much you trust other Italians in general?: (5) Trust them completely
(4) Trust them a little (3) Neither trust them, nor distrust them (2) Do
not trust them very much (1) Do not trust them at all." In the original
survey the numerical code of the response was in the reverse order.
Use of checks The survey asked household heads "Did you or some other member of the SHIW
household issue checks in the course of the year to settle transactions?"
Percent wealth in The survey asked household heads "What is the average amount of cash SHIW
cash held in your family?"
Percent wealth in In a typical survey, households are asked first to report ownership of the SHIW
stocks specific financial instrument and then to indicate the portfolio share, in
1989, or to report the asset bracket in a list of 14 possible brackets, in
1991, 1993, and 1995. In 1989 assets amounts are obtained combining
knowledge of the shares, of the value of financial wealth held in cash
and the fact that portfolio shares add up to 1. In 1991, 1993, and 1995,
assets amounts are imputed assuming that the household holds the
midpoint of the reported interval. It is clear from this procedure that
while stocks and mutual funds ownership only suffers from nonreporting,
their amounts is affected by imputation errors. For details on how
financial assets values are computed in the SHIW see Guiso and Tullio
Jappelli (2001).
Discouraged or turned The survey asked the following questions "During the year did you or a SHIW
down member of the household think of applying for a loan or a mortgage to a
bank or other financial intermediary, but then changed your mind on the
expectation that the application would have been turned down?" We
classify "yes" as "discouraged borrowers." The survey also asked
"During the year did you or a member of the household apply for a loan
or a mortgage to a bank or other financial intermediary and your
application was totally or partially turned down?" We classify answers
"yes totally" and "yes partially" as "turned down consumers."
Loans from friends The survey asked household heads "As of the end of the year did you have SHIW
and family debts outstanding towards friends or relatives not living with you? If yes,
what is their amount?" This information is used to compute the existence
and value of informal loans.

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
TABLE Al-Continued.
Variable Description Source

Judicial inefficiency
Mean ISTAT
number of years it takes to complete a first-degree trial by the courts
located in a province. It has been computed using courts-level data on
the length of trials and then averaging out across courts located in the
same province.
North/Center/South
Geographically we divide Italy in three regions. Provinces north of Our elaboration
Florence are located in the North, provinces between Florence and Rome
are located in the center, and provinces south of Rome are in the South.
We also use a finer partition of the territory into five macroareas: North
East, North West, Center, South, and Islands, according to ISTAT
definition.
Per capita GDP
GDP in the province ISTAT
in thousands of dollars divided by population in the
province.
Years of education Average number of schooling years calculated at the provincial level in ISTAT
1981.
Income/Wealth Income is the sum of the earnings of all members of the households that SHIW
worked for part or the whole year, pension income accruing to retired
members, capital income, and transfers. Wealth is the total of financial
and real assets net of household debt. The first is the sum of cash
balances, checking accounts, savings accounts, postal deposits,
government paper, corporate bonds, mutual funds, investment fund units,
and stocks. In 1989 total financial wealth is readily available. For other
years it must be estimated because the categories of financial assets
(except cash holdings) were provided in 15 bands; the average value
between the lower and the upper band was used in determining the level
of each asset. Real assets include investment real estate, business wealth,
primary residence, and the stock of durables. All the monetary variables
are deflated using the Consumer Price Index and expressed in dollars.
Age Household head age. SHIW
Education This variable is originally coded as: no education (0); completed SHIW
elementary school (5 years); completed junior high school (8 years);

completed high school (13 years); completed college (18 years); graduate
education (more than 20 years). The variable is coded according to the
values given in parentheses. For the highest class we assume a value of
20 years. It refers to the household head.
Married Indicator variable equal to one if the household head is married. SHIW
Male Indicator variable equal to one if the household head is a male. SHIW
Family size It includes all the individuals living in the house (adults and children) SHIW
Industry and job Industry dummies are a series of dummies for the industry where the SHIW
dummies household head works. Job dummies are a series of dummies for the
type of job (employee, professional, self-employed) held by the
household head.
Relative risk aversion Relative risk aversion is the product of the Arrow-Pratt measure of SHIW
absolute risk aversion and household's consumption. The Arrow-Pratt

measure of absolute risk aversion is obtained from a direct question to a
survey lottery where individuals report their willingness to pay for a
hypothetical risky security. Specifically, they are asked: "We would like
to ask you a hypothetical question that we would like you to answer as
if the situation was a real one. You are offered the opportunity of
acquiring a security permitting you, with the same probability, either to
gain 10 million lire or to lose all the capital invested. What is the most
that you are prepared to pay for this security?" Ten million lire
correspond to about USD 5,500. The respondent can answer in one of
following three ways: (a) declare the maximum amount he is willing to
pay to participate; (b) don't know; (c) unwilling to answer.
Crime Number of murders, robberies, and blackmails divided by population in ISTAT
1996.

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC
REFERENCES emy C. "Social Interaction and Stock Market

Participation." Journal of Finance, February
2004, 59(1), pp. 137-63.
Arrow, Kenneth. "Gifts and Exchanges." Philos-
ophy and Public Affairs, 1972, 1, pp. 343-62.
Ichino, Andrea and Maggi, Giovanni. "Work En-
vironment and Individual Background: Ex-
Banfield, Edward C. The moral basis of a back-
ward society. New York: Free Press, 1958.
plaining Regional Shirking Differentials in a
Besley, Timothy and Coate, Stephen. "Group
Large Italian Firm." Quarterly Journal of
Lending, Repayment Incentives and Social
Economics, August 2000, 115(3), pp. 1057-
Collateral." Journal of Development 90.
Eco-
nomics, February 1995, 46(1), pp. 1-18.
Knack, Stephen and Keefer, Philip. "Does Social
Bourdieu, Pierre. "The Forms of Capital," inHave an Economic Payoff? A Cross-
Capital
J. G. Richardson, ed., Handbook of theory
Country Investigation." Quarterly Journal of
and research for the sociology of education.
Economics, November 1997, 112(4), p. 1251.
New York: Greenwood Press, 1986, pp.Knack,
241- Stephen and Zack, Paul. "Trust and
58. Growth." Economic Journal, April 2001,
Brandolini, Andrea and Cannari, Luigi. "Meth- 111(470), pp. 295-321.
odological Appendix: The Bank of Italy's La Porta, Rafael; Lopez de Silanes, Florencio;
Survey of Household Income and Wealth," in Shleifer, Andrei and Vishny, Robert. "Legal
Albert Ando, Luigi Guiso, and Ignazio Visco, Determinants of External Finance." Journal
eds., Saving and the accumulation of wealth: of Finance, July 1997a, 52(3), pp. 1131-50.
Essays on Italian household and government . "Trust in Large Organizations." Amer-
saving behavior. Cambridge: Cambridge ican Economic Review, May 1997b, (Papers
University Press, 1994, pp. 369-86. and Proceedings), 87(2), pp. 333-38.
Case, Ann and Katz, Larry. "The Company You Merton, Robert. "Optimal Consumption and
Keep: The Effect of Family and Neighbor- Portfolio Rules in a Continuous-Time
hood on Disadvantaged Youth." National Bu- Model." Journal of Economic Theory, 1971
reau of Economic Research (Cambridge, 3, pp. 373-413.
MA) Working Paper No. 3805, 1991. Pagano, Marco; Panetta, Fabio and Zingales,
Coleman, James S. Foundations of social theory. Luigi. "Why Do Companies Go Public? An
Cambridge, MA: Harvard University Press, Empirical Analysis." Journal of Finance,
1990. February 1998, 53(1), pp. 27-64.
Fukuyama, Francis. Trust: The social virtues Portes, Alejandro. "Social Capital: Its Origins
and the creation of prosperity. New York: and Applications in Modem Sociology."
Free Press, 1995. Annual Review of Sociology, August 1998,
Guiso, Luigi and Jappelli, Tullio. "Household 24, pp. 1-24.
Portfolios in Italy," in Luigi Guiso, Michael Putnam, Robert D. Making democracy work:
Haliassos, and Tullio Jappelli, eds., House- Civic traditions in modern Italy. Princeton,
hold portfolios. Cambridge, MA: MIT Press, NJ: Princeton University Press, 1993.
2001, pp. 251-89. Solow, Robert M. "Trust: The Social Virtues and
Guiso, Luigi; Jappelli, Tullio and Pistaferri, Luigi. the Creation of Prosperity (Book Review)."
"An Empirical Analysis of Earnings and Em- New Republic, 1995, 213, pp. 36-40.
ployment Risk." Journal of Business and Spagnolo, Giancarlo. "Social Relations and Co-
Economic Statistics, April 2002, 20(2), pp. operation in Organizations." Journal of Eco-
241-53. nomic Behavior and Organization, January
Hong, Harrison; Kubik, Jeffrey D. and Stein, Jer-1999, 38(1), pp. 1-25.

132.174.251.250 on Sun, 12 Mar 2023 20:09:06 UTC

The Role of Social Capital in Financial Development

Uploaded by

Copyright:

Available Formats

The Role of Social Capital in Financial Development

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Role of Social Capital in Financial Development

Uploaded by

Copyright:

Available Formats

The Role of Social Capital in Financial Development

Author(s): Luigi Guiso, Paola Sapienza and Luigi Zingales

Stable URL: https://www.jstor.org/stable/3592941

This content downloaded from

By LUIGI GUISO, PAOLA SAPIENZA, AND LUIGI ZINGALES*

To identify the effect of social capital on financial development, we exploit social

In 1958 when Edward C. Banfield wrote developments in economic theory allow us to

This content downloaded from

are neither legal nor economic incentives to

This content downloaded from

This content downloaded from

2 The risk of the broker absconding is meant to capture

This content downloaded from

This content downloaded from

TABLE 1 SUMMARY STATISTICS AND CORRELATIONS

Panel A: Summary Statistics

Social capital 1 0.8 0.83 0.08 0.62 0.92 32,665

Social Social Participation Social Trust Judicial Per capita Average

This content downloaded from

FIGURE 1. TRUST ACROSS ITALIAN PROVINCES: PARTICIPATION IN REFERENDA

Note: Darker areas correspond to provinces with a higher participation in referenda.

This content downloaded from

Table 1 reports summary statistics for alleconomic

This content downloaded from

which is a function of the level of socialparatively

This content downloaded from

TABLE 2 EFFECT OF SOCIAL CAPITAL ON THE USE OF CHECKS

Social capital 1 0.5710*** 0.4265* 0.5552** 1.2584***

This content downloaded from

we think it is necessary to insert it into

This content downloaded from

This content downloaded from

TABLE 3-EFFECT OF SOCIAL CAPITAL ON PORTFOLIO SHARES

Panel A: Percentage of Wealth Invested in Cash

Panel B: Percentage of Wealth Invested in Stocks

This content downloaded from

Panel B: Percentage of Wealth Invested in Stocks-Continued.

Per capita GDP 0.0001 0.0001 -0.0013 -0.0004 -0.0020 -0.0003

Notes: In Panel A the dependent variable is the proportion of financial

of income squared and iswealth squared

This content downloaded from

on the proportion of wealth retained in by cash.

This content downloaded from

TABLE 4--EFFECT OF SOCIAL CAPITAL ON THE AVAILABILITY OF CONSUMER CREDIT

Social capital 1 -0.0588*** -0.0896*** -0.0986*** -0.1957***

Notes: The dependent variable is an indicator variable taking value one if

D. Availability of Credit to Consumers turned-dow

This content downloaded from

This content downloaded from

TABLE 5 EFFECT OF SOCIAL CAPITAL ON THE INFORMAL CREDIT MARKET

Notes: The dependent variable is an indicator variable that ta

IV. When Does Social Capital Matter

This content downloaded from

This content downloaded from

This content downloaded from

TABLE 6-SOCIAL CAPITAL AND LAW ENFORCEMENT

Social capital 1 0.4022 0.8537*** -0.7593*** 1l.0525*** 0.8600** 2.8714***

Social capital 1 -0.0010 -0.1338*** -0.0374 0.1543***

This content downloaded from

Social capital 1 0.5710*** 0.4265* 0.5552 1.2584*

Social capital 1 -0.0588* -0.0896* -0.0986* -0.1957*

Social capital 1 0.4022 0.8537* -0.7593* 1l.0525* 0.8600 2.8714***

Social capital 1 -0.0010 -0.1338* -0.0374 0.1543*

Social capital 1 0.7131* 0.0135 -1.0348* -0.3442* 1.8479* 1.5451***

Social capital 0.1797 -0.1750* 0.0273 -0.0321 -0.0606