American Economic Association

Trust in Large Organizations

Author(s): Rafael La Porta, Florencio Lopez-de-Silanes, Andrei Shleifer and Robert W. Vishny
Source: The American Economic Review, Vol. 87, No. 2, Papers and Proceedings of the
Hundred and Fourth Annual Meeting of the American Economic Association (May, 1997), pp.
333-338
Published by: American Economic Association
Stable URL: http://www.jstor.org/stable/2950941 .
Accessed: 02/04/2013 23:02

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
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 support@jstor.org.

American Economic Association is collaborating with JSTOR to digitize, preserve and extend access to The
American Economic Review.

http://www.jstor.org

This content downloaded from 190.193.83.6 on Tue, 2 Apr 2013 23:02:49 PM

All use subject to JSTOR Terms and Conditions
 Trust in Large Organizations

By RAFAEL LA PORTA, FLORENCIOLOPEZ-DE-SILANES,

ANDREI SHLEIFER,AND ROBERT W. VISHNY*

Several recent studies, including those in ciation between trust and the dominance of a
Diego Gambetta (1988), as well as studies strong hierarchicalreligion in a country, most
by James Coleman (1990), Robert Putnam notably Catholicism.
(1993), and Francis Fukuyama (1995), ar-
gue that trust or social capital determines the I. Argument
performance of a society's institutions.
These studies view trust or social capital as Economists have developed two views of
a propensity of people in a society to coop- trust as a tendency to cooperate. One view,
erate to produce socially efficient outcomes rooted in repeatedgame theory, holds thattrust
and to avoid inefficient noncooperative traps is a priorthat an opponent is cooperativerather
such as that in the prisoner's dilemma. than fully rational (e.g., plays only tit-for-tat
Putnam (1993), for example, examines so- in a repeated prisoner's dilemma). A higher
cial capital as a determinant of the perfor- prior in a repeatedprisoner's dilemma leads to
mance of local governments across Italian a greater likelihood and duration of coopera-
regions. He demonstrates that the Italian tion (David Kreps et al., 1982). Another view,
regions in which the public actively partici- rooted in economic experiments, holds that
pates in civic activities (viewed as manifes- people cooperate even in one-shot encounters,
tations of a high tendency to cooperate) are such as the dictator game or the ultimatum
also the regions in which local governments game (Colin Camerer and Richard Thaler,
exhibit higher objective measures of perfor- 1995). These experiments suggest that people
mance, such as the delivery of public goods. expect certain fair or cooperative behavior of
Fukuyama (1995) argues further that high their opponents even when they do not expect
trust among citizens accounts for the supe- to see them again. Both of these views sug-
rior performance of all institutions in a so- gest that higher trustbetween people in a pop-
ciety, including firms. ulation should be associated with greater
In this paper, we provide an overview of the cooperation.
existing theory of trust, develop some of its These views of trust share an importantim-
implications, and test them on a cross section plication, namely, that trust should be more
of countries. We find a striking confirmation essential for ensuring cooperation between
of the theory in the data. We also ask: what strangers,or people who encounter each other
are the forces that encourage the formation of infrequently, than for supporting cooperation
trust? According to Putnam (1993), hierar- among people who interact frequently and re-
chical religion discourages "horizontal" ties peatedly. In the latter situations, such as fam-
between people and hence the formation of ilies or partnerships, reputations and ample
trust. Indeed, we find a strong negative asso- opportunities for future punishment would
support cooperation even with low levels of
trust. This implies that trust is most needed to
support cooperation in large organizations,
* La Porta,Lopez-de-Silanes, and where members interact with each other only
Shleifer:Department
of Economics, Harvard University, Cambridge, MA infrequently because they are only rarely in-
02138; Vishny: GraduateSchool of Business, University volved in joint production.Take, for example,
of Chicago, 1101 East 59th St., Chicago, IL 60637. We administrative interactions between members
are grateful to Olivier Blanchard,Edward Glaeser, Henry
Hansmann, Larry Katz, and Richard Thaler for helpful of different departmentsin a university, or in-
comments and to Andrew Prihodko and Magdalena teragency task forces in the government. Here
Lopez-Morton for research assistance. cooperation is less sustainable without trust
333

This content downloaded from 190.193.83.6 on Tue, 2 Apr 2013 23:02:49 PM

All use subject to JSTOR Terms and Conditions
334 AEA PAPERS AND PROCEEDINGS MAY 1997

because interactions are too few to allow rep- people answering yes is our measure of trust
utations to develop. within a country (see also Stephen Knack and
One such large organization is the govern- Philip Keefer, 1996). The correlation across
ment, where bureaucrats must cooperate countries between trust in the 1980's and in
with a large number of other bureaucrats the 1990's is 0.91, so we use the later, more
whom they encounter only infrequently, as complete, data. The highest-trustcountries are
well as with private citizens they may never in Scandinavia, where almost two-thirdsof the
see again, to produce "public goods." Sig- respondents believe that strangers can be
nificant trust is then needed to ensure coop- trusted;many of the lowest-trust countries are
eration. Local governments in Italy might in Latin America.
perform better in high-trust regions because Here we briefly summarize our variables;
trust enables individual bureaucrats to co- Table 1 provides details. For government ef-
operate better with each other and with pri- fectiveness we use (subjective) estimates from
vate citizens, making government more investor surveys of the efficiency of the judi-
effective. cial system, corruption, bureaucraticquality,
Civic groups or associations, where par- and tax compliance (a proxy for effectiveness
ticipation is largely voluntary and success of the tax authorities). For civic participation
depends on many people cooperating, may we use the extent of participationin civic ac-
also rely on trust for their success. Putnam tivities and in professional associations. To
(1993) actually measures social capital by measure the relative success of large firms, we
participation in civic groups and associa- use total sales of the largest 20 publicly traded
tions, even though participation must itself firms (by sales) in a country relative to its
be a consequence of some underlying beliefs GNP. Some of our measures of social success
about the behavior of other people in the proxy for the effectiveness of government
society. only; others reflect the success of other insti-
Finally, corporations are also large organ- tutions in the society as well. We look at ed-
izations that would benefit from trust among ucation (high-school graduatesrelative to the
their employees. Fukuyama (1995) stresses relevant population and educational adequacy
the need for cooperation between strangers as estimated by a business group), health (log-
for the success of large firms, and the de- arithmof infant mortality rate), infrastructure
pendence of such cooperation on trust. He (an estimate of its quality by one business
contrasts large public firms in high-trust group and an estimate of its adequacy by an-
countries to smaller family firms that prevail other), and two general measures of social
in low-trust societies. success: per capita GDP growth between 1970
and 1993 and (the logarithmof) inflation over
II. Evidence the same period. Our goal is to establish the
robustness of the results through the use of
We test the hypotheses developed above on multiple variables and data sources; we have
a cross section of countries. We are interested looked at many other variables as well, with
in the effect of trust on the performance of similar results.
large organizations,measuredhere by govern- Table 2 presents the regressions of our mea-
ment effectiveness, participation in civic or- sures of performanceof large organizationson
ganizations, size of the largest firmsrelative to trust, controlling for the log of 1994 per capita
GNP, and the performance of a society more GNP. In most regressions, we have fewer than
generally. Our measure of trust comes from 40 observations because we do not have de-
the World Values Survey, which in the early pendent variables for socialist countries. Con-
1980's and again in the early 1990's surveyed trolling for per capita GNP reduces the effect
1,000 randomly selected people in each of 40 of trust,since trustis higher in richercountries.
countries. One of the questions was: "Gener- This may cause the effect of trust to be under-
ally speaking, would you say that most people estimated if trust is an input in the production
can be trusted or that you can't be too careful of wealth. In other (unreported) specifications,
in dealing with people?" The percentage of we also control for inequality without much

This content downloaded from 190.193.83.6 on Tue, 2 Apr 2013 23:02:49 PM

All use subject to JSTOR Terms and Conditions
VOL. 87 NO. 2 INTERACTIONOF ECONOMICINSTITUTIONSAND THEORY 335

TABLE 1-DESCRIPTION OF THE VARIABLES USED TABLE 1-Continued.

IN TABLES 2 AND 3
Variable Definition

Variable Definition GDP growth Average annual growth in per capita GDP for the
period 1970-1993. Source: WDR95.
Trust in people Percentage of respondentswho answered that most Log GNP per Logarithmof the GNP per capita expressed in dollars
people can be trustedwhen asked: "Generally capita of 1994 unless otherwise noted. Source: World
speaking, would you say that most people can be DevelopmnentReport 1996.
trustedor that you can't be too careful in dealing Trust in family Rating based on respondents' answers to how much
with people?" Source: World Values Survey 1990- they trust their families. Scale from 0 to 4. The
93 (WVS). highest (lowest) rating is awardedwhen respondents
Efficiency of the Assessment of the "efficiency and integrity of the legal manifest that they trust (distrust) their families.
judiciary environment as it affects business, particularly Source: WVS.
foreign firms." Average between 1980 and 1983. Hierarchical Percentage of the population of each country that are
Scale from 0 to 10, with lower scores indicating religion Roman Catholic, Eastem Orthodox, or Muslim.
lower efficiency levels. Source: Business Sources: WorldmarkEncyclopedia of the Nations
International Corporation ( 1984). 1995, Statistical Abstract of the World 1994.
Corruption Low ratings if "high government officials are likely to Ethnolinguistic Probabilitythat two randomly selected persons from a
demand special payments and illegal payments are fractionalization given country will not belong to the same
generally expected throughoutlower levels of ethnolinguistic group in 1960. Source: World
government in the form of bribes connected with Handbook of Political and Social Indicators.
import and export licenses, exchange controls, tax
assessment, policy protection, or loans." Scale from
0 to 10. Average of the months of April and October
of the monthly index between 1982 and 1995.
Source: InternationalCountryRisk Guide (ICRG).
Bureaucratic High scores indicate "autonomy from political
quality pressure" and "strength and expertise to govern effect. We interpret the coefficients using a
without drastic changes in policy or interruptionsin
government services." Scale from 0 to 10, with one-standard-deviationchange in trust, about
higher scores for greaterefficiency. Average of the 15 percentage points.
months of April and October of the monthly index
between 1982 and 1995. Source: ICRG. The effects of trust on performanceare both
Tax compliance Assessment of the level of tax compliance. Scale from
0 to 6, where higher scores indicate higher
statistically significant and quantitatively
compliance. Source: The Global Competitiveniess large. Holding per capita GNP constant, a
Report 1996 (GCR).
Civic participation Percentageof civic activities in which an average
standard-deviationincrease in trust raises ju-
individual participates.The activities included are: dicial efficiency by 0.7, the anticorruption
(i) social-welfare services for elderly and deprived,
(ii) education, art, and culturalactivities, (iii) local score by 0.3, bureaucraticquality by 0.3, and
community affairs, (iv) conservation, environment, tax compliance by 0.3 of a standarddeviation.
ecology, (v) work with youth, (vi) sports or
recreation,and (vii) voluntaryassociations for Putnam's (1993) results for Italy appear to
Participationin
health. Source: WVS.
Percentage of respondentswho answered positively
be confirmed worldwide. Furthermore, a
professional when asked if they belonged to professional standard-deviationincrease in trustraises par-
associations associations. Source: WVS.
Sales' top 20/ The ratio of sales generated by the top 20 publicly
ticipation in civic activities by 0.7 and
GNP traded firms to GNP for 1994. Firms within a participation in professional associations by
country are ranked by sales. Source: WorldScope
Global 1996 data base. one standarddeviation. The effect of trust on
Infrastructure Assessment of the "facilities for and ease of large firms' share of the economy is also large:
quality communication between headquartersand the
operation, and within the country," as well as the a one-standard-deviation increase in trust
quality of the transportation.Average data for the
years 1972-1995. Scale from 0 to 10 with higher
raises that share by 7 percentagepoints, or half
scores for superiorquality. Source: BERI's of a standarddeviation. These results support
Operations Risk Index.
Adequacy of Average of five scores measuring the extent to which a
Fukuyama's (1995) argumentthat trust facil-
infrastructure country's infrastructuremeets business needs in each itates all large-scale activities, not just those of
of the following areas: (i) roads, (ii) air transport,
(iii) ports, (iv) telecommunications, and (v) power the government.
supply. Scale from 0 to 6, where higher score'sare Indeed, Fukuyama goes further and argues
for a superior infrastructure.Source: GCR.
Log infant Logarithmof the number of deaths of infants under that, for firms in particular,trust replaces an-
mortality one year of age per one thousand live births for other mechanism of cooperation-the family.
1993 or the most recent year available. Source:
Health-For-All Global Indicators Database. He believes that family strengthis detrimental
Completed high Percentageof the 1985 male population aged 25 and
school over that has completed high school. Source: Robert
to the growth of firms. We can actually test
Barroand Jong-Wha Lee ( 1994). this hypothesis since the World Values Survey
Adequacy of Assessment of the extent to which the educational
educational system meets the needs of a competitive economy. asks respondents if they trust their families.
system Score from 0 to 6, where higher scores are for a When we run the relative share of the top 20
superioreducational system. Source: GCR.
Log inflation Logarithmof the geometric average annual growth rate firms on both a measure of trust in strangers
of the implicit price deflatorfor the time period
1970-1993. Source: WorldDevelopment Report
and a measure of trustin family, the coefficient
1995 (WDR95). on trust in people is 0.654 (t = 4.1), and the

This content downloaded from 190.193.83.6 on Tue, 2 Apr 2013 23:02:49 PM

All use subject to JSTOR Terms and Conditions
336 AEA PAPERS AND PROCEEDINGS MAY 1997

TABLE 2-TRUST IN PEOPLE AND PERFORMANCE and school adequacy by one-third of a stan-
dard deviation. Trust is also associated with
Independentvariables
lower inflation and weakly associated with a
Log GNP Trust in Adjusted
Dependent variable per capita people Intercept R2 [NJ
higher per capita GNP growth (about 0.3 per-
GovernmentEfficiency
cent per annumper standarddeviation increase
in trust). This result for growth was also ob-
Efficiency of the 0.2959 8.2093** 2.2769 0.6343
judiciary (0.2213) (1.3652) (1.7766) [271 tained by Knack and Keefer (1996). In sum,
Corruption 0.9214**
(0.1022)
4.8068**
(0.7509)
-2.3608**
(0.9050)
0.7316 trust enhances economic performance across
[331
Bureaucraticquality 1.1596** 3.9797** -4.0842* 0.6806 countries.
(0.1927) (1 .3544) (1.6763) [33]
Tax compliance 0.3595** 1.7330** -0.9124 0.3540
(0.0913) (0.5840) (0.7873) [32] III. WhereDoes TrustComeFrom?
Participation

Civic participation 0.0127** 0. 1224** -0.0921 ** 0.4614

Trust may not be truly exogenous; it may
(0.0038) (0.0329) (0.0308) [33] increase with good past performance of a so-
Participationin -0.0072 0.3056** 0.0330 0.5492
professional (0.0099) (0.0669) (0.0730) [331
ciety's institutions. According to Putnam
associations (1993), trust is a habit formed during a
Large Organizations centuries-long history of "horizontalnetworks
Sales' top 20/GNP 0.0103 0.4927** -0.0374 0.2433
of association" between people, covering both
(0.0325) (0.1657) (0.2798) [261 commercial and civic activities. Putnam ar-
Social Efficiency
gues that the independent city-states of
Northern Italy encouraged the formation of
Infrastructure 1.0269** 2.3261 ** -3.7162** 0.6783
quality (0.1413) (0.7970) (1.2331) [33] such horizontal networks, in contrast to the
Adequacy of
infrastructure
0.5943** 1.251 1** -1.6559**
(0.5837)
0.7222
[32]
more authoritarian political regimes of the
(0.0604) (0.4200)
Log infant mortality -0.4598** -1.0283* 6.9682** 0.7141 South. Can we measure something even more
rate (0.0484) (0.5176) (0.4495) [401 basic than trust?
Completed high 1.2884** 10.9714** -7.4405* 0.3474
school (0.4416) (3.4633) (3.5336) [29] Putnam ( 1993 ) argues that the Catholic
Adequacy of 0.2200** 1.2334* 0.8525 0.2107
education system (0.0858) (0.6771) (0.7736) [321
Church, by imposing a hierarchical structure
Log inflation 0.0371 -3.4128** 3.1306** 0.2059 on the society, often in symbiosis with the
(0.0787) (1.1502) (0.6494) [37]
GDP growth -0.2738t 2.0266t 3.5847** 0.0072 state, has discouraged the formation of trust:
(0.1548) (1.2152) (1.3625) [391 "Vertical bonds of authority are more char-
Notes: Ordinaryleast-squareregressions of the cross section of 40 countries.
acteristic of the Italian Churchthan horizontal
There are 14 dependent variables classified in four differentgroups including bonds of fellowship" (p. 107). His argument
(i) Government Efficiency, (ii) Participation,(iii) Large Organizations, and
(iv) Social Efficiency Coefficients are shown, and HalbertWhite (1980) cor- can be applied more generally to any domi-
rected standarderrors are given in parentheses underneath.The number of nant, hierarchical, organized religion in a
observations is given in brackets.
Statistically significant at the 10-percent level. country and hence can be tested empirically
* Statistically significant at the 5-percent level.
** Statistically significant at the I-percent level.
across countries.
Specifically, for every country, we consider
the percentage of the population belonging to
a hierarchical religion, defined as Catholic,
coefficient on trust in family is -0.563 (t = Eastern Orthodox, or Muslim. The mean of
-3.1). Consistent with Fukuyama's ar- this variable in the sample is 55 percent, and
gument, strong family ties are bad for the its correlationwith trustis a remarkable-0.61
development of large firms. (see Fig. 1). This correlation (and all of the
The last panel of Table 2 presents the results following results) is driven mostly by the cor-
for social outcomes. Trust has a relatively relation of -0.47 between percentageCatholic
small but significant effect on infrastructure and trust, although predominantly Moslem
quality and adequacy, a significant effect on and Eastern Orthodox countries have very
infant mortality, and a larger effect on the low trust as well. In Table 3, we use this
measures of educational achievement. A one- hierarchical-religion variable as an indepen-
standard-deviationincrease in trust raises the dent variable to explain organizational
percentage of high-school graduates in the performance. Holding per capita income con-
populationby one-half of a standarddeviation, stant, countries with more dominanthierarchi-

This content downloaded from 190.193.83.6 on Tue, 2 Apr 2013 23:02:49 PM

All use subject to JSTOR Terms and Conditions
 VOL. 87 NO. 2 INTERACTIONOF ECONOMICINSTITUTIONSAND THEORY 337

0.661- SIY TABLE 3-RELIGION AND PERFORMANCE

FNLO
CHNA
DNMK
Independentvariables
NTH.L CNOA
US A
Log GNP Hierarchical Adjusted
U3) GALD Dependent variable per capita Religion Intercept R2 [N]
0. ICPA UK SWTZ
a) GovernmentEfficiency
Q- GERM AUSS

<OGS INA8GMX"

1<TH AUST BKGP

Efficiency of the 0.7420** -0.0233** 2.4936 0.5245
SAFA
:3
ESTO CFC, judiciary (0.2357) (0.0067) (2.4613) [271
HUNG 8ELA Corruption 1.0740** -0.0148** -1.1331 0.7025
LATV
(0.1171) (0.0051) (1.2028) [33]
SLOV RMNA Bureaucraticquality 1.2376** -0.0214** -2.1445 0.7583
(0.1493) (0.0054) (1.5465) [331
Tax compliance 0.3985** -0.0088** -0.1415 0.4335
0.065 UAZL (0.0767) (0.0029) (0.6727) [321
0.3 99.8 Participation
Hierarchical Religion, Percentage of Population
Civic participation 0.0164** -0.0003** -0.0589* 0.4106
(0.0035) (0.0001) (0.0274) [33]
FIGURE 1.
TRUST IN PEOPLE PLOTTED AGAINST Participationin 0.0002 -0.0010** 0.1393 0.5256
THE PERCENTAGE OF THE POPULATION BELONGING professional (0.0105) (0.0002) (0.1104) [33]
associations
TO A HIERARCHICAL RELIGION
Large Organizations

Sales' top 20/GNP 0.0370 -0.0020** 0.0077 0.3387

(0.0263) (0.0006) (0.2668) [26]
Social Efficiency

Infrastructure 1.0725** -0.0172** -2.3035* 0.7835

cal religions have less efficient judiciaries, quality (0.0955) (0.0039) (0.8967) [33]
greater corruption, lower-quality bureaucra- Adequacy of 0.6252** -0.0057* -1.1572* 0.7480
infrastructure (0.0550) (0.0021) (0.5270) [32]
cies, higher rates of tax evasion, lower rates of Log infant -0.5044** 0.0016* 6.9040** 0.6919
participation in civic activities and profes- mortality rate (0.0473)
1.7590**
(0.0018)
-0.0409**
(0.4794)
-5.4915
[40]
0.3814
Completed high
sional associations, a lower level of impor- school (0.4573) (0.0142) (4.3376) [29]
-0.0016 0.1597
tance of large firms in the economy, inferior Adequacy of
education system
0.2680**
(0.0820) (0.0028)
0.9735
(0.7854) [32]
infrastructures,and higher inflation. The re- Log inflation -0.0915 0.0087* 2.5343** 0.0991
(0.0784) (0.0035) (0.7381) [37]
sults for infant mortality,educational achieve- GDP growth -0.1944 -0.0030 3.8230* -0.0206
ment, and growth are less clear-cut. Still, the (0.1622) (0.0069) (1.6884) [39]
evidence that hierarchicalreligions are bad for Notes: There are 14 dependent variables classified in four different groups
the performance of large organizations is including (i) Government Efficiency, (ii) Participation,(iii) Large Organiza-
tions, and (iv) Social. We reportcoefficients for heteroscedasticity-corrected
strong. We have also run two-stage specifica- OLS (White, 1980). Standarderrors are shown in parentheses. The number
tions, in which hierarchicalreligion is used as of observations is given in brackets.
* Statistically significant at the 5-percent level.
an instrumentfor trust. The results are similar ** Statistically significant at the 1-percentlevel.
to those in Table 2 in both magnitude and sta-
tistical significance.
This evidence suggests that hierarchicalre-
ligion and distrust may both reflect some un- IV. Conclusion
derlying basic "factor" in a society that is
detrimentalto the performanceof large organ- Trust promotes cooperation, especially in
izations. This factor may reflect dysfunctional large organizations. Data on government per-
institutions in a society, but if so, this is largely formance, participation in civic and profes-
a long-term disfunctionality associated in part sional societies, importance of large firms,
with a hierarchical religion (and not just with and overall performance of different societies
recent events). Interestingly, this factor does support this hypothesis. Furthermore,trust is
not reflect the ethnic heterogeneity in a soci- lower in countries with dominant hierarchi-
ety, which might be viewed as a source of dis- cal religions, which may have deterred the
trust: the correlation between trust and a formation of "horizontal networks of coop-
standardmeasure of ethnolinguistic heteroge- eration" among people. Despite economists'
neity is only -0.12, and the inclusion of that skepticism (see Robert Solow, 1995), theories
measure in the regressions in Table 2 does not of trust hold up remarkablywell when tested
change the importance of trust. on a cross section of countries.

This content downloaded from 190.193.83.6 on Tue, 2 Apr 2013 23:02:49 PM

All use subject to JSTOR Terms and Conditions
338 AEA PAPERS AND PROCEEDINGS MAY 1997

REFERENCES Cross-Country Investigation." Mimeo,

University of Maryland, 1996.
Barro,Robertand Lee, Jong-Wha.Unpublished Kreps,David;Milgrom,Paul; Roberts,John and
data set for a panel of 138 countries, Har- Wilson,Robert."Rational Cooperationin the
vard University, 1994. Finitely Repeated Prisoner's Dilemma."
BERI'soperationsriskindex.Business environ- Journal of Economic Theory, August 1982,
mental risk intelligence indicators obtained 27(2), pp. 245-52.
by the Center for Institutional Reform and Putnam, Robert. Making democracy work:
the Informal Sector (IRIS), University of Civic traditions in modern Italy. Princeton,
Maryland,various years. NJ: Princeton University Press, 1993.
BusinessInternationalCorporation.Introduction Solow, Robert. "But Verify," Review of
to the Country Assessment Service. New Fukuyama. New Republic, 11 September
York: Business International Corporation, 1995, p. 36.
1984. Statisticalabstractof the world1994. New York:
Camerer,Colinand Thaler,RichardH. "Anom- Gale Research, 1994.
alies: Ultimatums, Dictators, and Man- White, Halbert. "A Heteroskedasticity-
ners." Journal of Economic Perspectives, Consistent Covariance Matrix Estimator
Spring 1995, 9(2), pp. 209-20. and a Direct Test for Heteroskedasticity."
Coleman,James.Foundations of social theory. Econometrica, May 1980, 48(4), pp. 817-
Cambridge,MA: HarvardUniversity Press, 38.
1990. Worlddevelopmentreport1995. New York: Ox-
Fukuyama, Francis. Trust. New York: Free ford University Press, 1995.
Press, 1995. Worlddevelopmentreport1996. New York: Ox-
Gambetta,Diego,ed. Trust:Making and break- ford University Press, 1996.
ing cooperative relations. Cambridge: Worldhandbookof politicaland social indicators
Blackwell, 1988. III, 1948-1982(ICPSR7761).Ann Arbor,MI:
Theglobalcompetititivereport1996.Geneva, Swit- Inter-university Consortium for Political
zerland:World Economic Forum, 1996. and Social Research, 1983.
Health-for-Allglobal indicators database. Ge- Worldvalue survey, 1981-1984 and 1990-1993
neva, Switzerland: World Health Organi- (ICPSR6160).Ann Arbor,MI: Inter-university
zation, 1996. Consortium for Political and Social Re-
Internationalcountryrisk guide. East Syracuse, search, 1994.
NY: Political Risk Services, InstitutionalRe- Worldmark encyclopediaof the nations1995, 8th
form and InformationalSector,variousyears. Ed. Detroit, MI: Gale Research, 1996.
Knack,StephenandKeefer,Philip."Does Social WorldScope global 1996. Bridgeport, CT:
Capital Have an Economic Payoff? A WorldScope Disclosure Partners, 1996.

This content downloaded from 190.193.83.6 on Tue, 2 Apr 2013 23:02:49 PM

All use subject to JSTOR Terms and Conditions