Courts and Banks:

Effects of Judicial Enforcement on Credit Markets

Tullio Jappelli
CSEF, Universit di Salerno, and CEPR
Marco Pagano
CSEF, Universit di Salerno, and CEPR
Magda Bianco
Banca dItalia
April 2002
Abstract
The cost of enforcing contracts is a key determinant of market performance. We document
this point with reference to the credit market in a model of opportunistic debtors and
inefficient courts. According to the model, improvements in judicial efficiency should reduce
credit rationing and increase lending, with an ambiguous effect on interest rates that depends
on banking competition and on the type of judicial reform. These predictions are supported by
panel data on Italian provinces and by cross-country evidence. In Italian provinces with longer
trials or large backlogs of pending trials, credit is less widely available. International evidence
also shows that the depth of mortgage markets is inversely related to the costs of mortgage
foreclosure and other proxies for judicial inefficiency.
Keywords: enforcement, judicial efficiency, credit market, interest rates
JEL classification: G2, K4
Acknowledgements: We are grateful for helpful comments from seminar participants at the
Banque de France, European University Institute, the European Central Bank and at the
University of Naples, and particularly to Olivier de Bondt, Francesco Giavazzi, Reint Gropp,
Andrea Ichino and Lucio Scandizzo. This research has been supported by grants awarded by
the Fondation Banque de France, the Italian Ministry of Education, University and Research
(MURST) and the European Commission (Research Training Network on Understanding
Financial Architecture: Legal and Political Frameworks and Economic Efficiency). The views
expressed here are those of the authors and do not involve the responsibility of the Banca
dItalia.
1
May you have lawsuits - and win them.
Old gypsy curse
1
1. Introduction
A borrower may default on a loan because he is unable (accidental default) or because,
though potentially solvent, he is unwilling to repay (strategic default). Besides being
intrinsically different, inability and unwillingness to repay depend on totally different factors.
A borrower is unable to repay if his project fails, which may in turn depend on bad luck,
incompetence, poor effort in managing the project, or a combination of all three factors.
A solvent borrower may be unwilling to repay if the gain from defaulting is greater
than the perceived cost of the presumed sanctions. The perceived cost of these sanctions does
not depend only on the lenders willingness to inflict them, but on the entire set of
institutional arrangements governing the credit market. The law and its enforcement by the
judiciary are central to these arrangements. Historically, countries have developed different
legal systems, which feature varying degrees of protection of creditors rights. But even
countries with similar legal rules may enforce them to a differing extent, depending on the
efficiency and honesty of their judiciary. And even within the same country, the efficiency of
courts can vary a great deal, depending on the allocation of resources and the geographical
distribution of the demand for contract enforcement.
By affecting the borrowers future willingness to pay, these features help determine the
ex ante willingness of creditors to extend loans, and the terms they will ask. By the same
token, they determine the effectiveness of credit markets in intermediating and allocating
saving among alternative users.
This paper explores the impact of the judicial enforcement of debt contracts on the
amount of lending, interest rates and default rates theoretically and empirically. We present a
model of opportunistic debtors and inefficient courts. Judicial efficiency is measured by the
fraction of inside or outside collateral that lenders can expect to recover from an insolvent
borrower at the end of a trial. According to the model, an improvement in judicial efficiency
unambiguously increases aggregate lending, by opening the credit market to borrowers with

1
This double curse about the slowness of trials and the difficulty of obtaining damages once they are awarded is
drawn from the Financial Times, Weekend December 12/13 1998, p. 3.
2
little collateral. The impact of judicial efficiency on the average interest rate is ambiguous, in
that this depends on the structure of the credit market (competitive or monopolistic) and on
the specific judicial reform (improvement in the recovery of inside or outside collateral).
We then test these predictions empirically, using two data sets. The first is a specially
designed Italian panel on interest rates, lending, overdrafts, default rates, and indicators of
judicial efficiency in each province. The second is a cross-country sample of mortgage
lending, downpayment ratios and interest rates, plus measures of the cost and length of
foreclosure procedures.
The evidence from both data sets is that judicial enforcement is important to the
performance of credit markets. Our findings are that judicial efficiency correlates positively
with the volume of lending and negatively with proxies for credit rationing. These results
obtain also when in our panel data estimates we control for unobserved heterogeneity among
judicial districts via fixed effects. The correlation with average interest rates and default rates
is ambiguous, in line with the prediction of the model.
With the help of a simple illustrative model, in Section 2 we discuss the theoretical
channels through which judicial efficiency can affect credit market performance. In Section 3
we present our province-level data and the corresponding regression results. Section 4 gives
the international evidence on mortgage loan markets. Section 5 concludes.
2. A Model of Judicial Enforcement and Credit Markets
The key function of courts in credit relationships is to force solvent borrowers to repay when
they fail to do so spontaneously. Hence poor judicial enforcement will increase opportunistic
behavior on the part of borrowers: anticipating that creditors will be unable to recover their
loans easily and cheaply via the courts, borrowers will be tempted to default. Lenders respond
by reducing the availability of credit.
We illustrate how judicial inefficiency affects credit market performance in a model of
risk-neutral banks facing a continuum of potential borrowers. Each borrower i has no liquid
wealth but owns illiquid collateral
i
C . He can invest in a project requiring a loan of size
i
L ,
3
so that his collateral-loan ratio is
i i i
L C c / . Projects succeed with common probability p and
fail with probability p 1 . All successful projects yield + 1 per unit invested, and failed
projects yield zero. All projects have positive net present value (NPV), that is, their expected
profitability exceeds the banks cost of raising funds, r :
r p + > + 1 ) 1 ( .
Since r is also the opportunity cost of capital for entrepreneurs, all of them would like to
undertake their projects.
Banks can observe whether projects succeed or fail, so that there is no asymmetric
information.
2
In either case the borrower can dispute the banks claim. In case of dispute, the
bank can attempt to recover the loan in court. But it will recover only a fraction
p
of the
projects revenue
3
and a fraction
c
of the collateral. The parameters
p
and
c
can be
regarded as indicators of judicial efficiency. Both range from 0 (no enforcement) to 1 (perfect
enforcement).
There are two possible interpretations of this assumption. First, by disputing the
repayment and forcing the lender to go to court, the borrower retains a fraction of the loan
i c p
c ) 1 ( ) 1 )( 1 ( + + in case of success, and
i c
c ) 1 ( in case of failure. Since he can
pocket part of the firms revenue or consume part of the collateral, he has a clear incentive for
opportunistic behavior. He will always dispute the lender's claim, whether the project has
succeeded or failed.
A second interpretation is that these resources are not retained by the borrower, but
dissipated by the judicial process itself (legal fees, mismanagement of the company during the
trial, bribes taken by corrupt officials, etc.). In this interpretation, judicial costs effectively
operate as a tax on credit transactions. In principle, this tax can be avoided by settling out of
court, two parties having to agree on how to split the resources that they would have otherwise
wasted. If judicial costs are borne entirely by the lender, the borrower will make a take-it-or-
leave-it offer to repay
i c P
c + + ) 1 ( per dollar lent in case of success, and
i c
c in case of

2
The model can also effectively capture the case where the lender cannot observe the outcome of the project. In
this case, the borrower will always claim that the project has failed. Anticipating this, the lender will extend
credit only if repayment is guaranteed by collateral. In the model, this case would obtain with
p
=0.
3
The subscript p stands for project, since in this case the project itself acts as inside collateral.
4
failure. The lender will be indifferent between accepting this offer and taking the borrower to
court. In this case, the borrower retains the entire cost of the trial, and the two alternative
interpretations lead exactly to the same outcome. If judicial costs are more evenly distributed
between the two parties, the borrower could keep only part of the cost of the trial. Even so, he
will generally have an incentive to dispute the amount owed and thereby extract that portion
from the lender.
4
In short, borrowers reckon that lenders will be able to recover at most
i c p
c + + ) 1 (
per unit lent in case of success, and
i c
c otherwise. Thus the lending rate charged to borrower
i,
i
r , cannot exceed the limit:
i c p i
c r + + + ) 1 ( 1 . (1)
All banks know the success probability p, the projects profitability , the judicial
efficiency parameters
p
and
c
, and the individual borrowers collateral-loan ratio c
i
.
2.1 Competitive Banks
In equilibrium, expected profits are zero, so that the cost of funds equals the expected return
per unit lent to borrower i:
] , 1 min[ ) 1 ( ) 1 ( 1
i c i i
c r p r p r + + + = + . (2)
The last term states that when the project fails the lender recovers only a fraction of the
collateral if this falls short of the principal plus interest due. Equation (2) defines the break-
even interest rate
i
r

charged to each borrower:
) , 1 min(
1 1
1
i c i i
c r
p
p
p
r
r +

+
= + for
min
c c
i
, (3)
where:

4
If lenders bear only a fraction of judicial costs, the borrowers take-it-or-leave-it offer will be accordingly
reduced to [ ] [ ]
i c p
c ) 1 ( 1 ) 1 ( ) 1 ( 1 + + in case of success and to [ ]
i c
c ) 1 ( 1 in case of failure. The
feasibility condition (1) and all subsequent expressions must be redefined accordingly. All the comparative
statics concerning an improvement in judicial efficiency are qualitatively unchanged.
5
c
p
c
p
r
c

) 1 (
1
min
+

+
= . (4)
The minimum level of collateral in equation (4) is obtained by substituting (1) (taken with
equality) into equation (3). Banks do not finance entrepreneurs with collateral-loan ratio
below
min
c , even though with internal financing their projects would be profitable. Thus
min
c defines the region of credit rationing. This is due only to judicial inefficiency: with
efficient courts ( 1 = =
p c
) all entrepreneurs would have access to credit.
5
The zero-profit condition (3) defines two lending regions. If
i i c
r c + > 1 , collateral is
large enough that loans are safe and competition equates the lending rate to the cost of capital.
Setting r r
i
= in equation (3) yields the level of collateral above which this happens:
c
r
c

+
=
1
. (5)
In the second region,
i i c
r c + < 1 or equivalently c c
i
< : collateral is not sufficient to
shield the bank completely from loss if the project fails. To break even, the bank must offset
this expected loss with a higher interest rate in case of success: from the standpoint of the
bank, collateral and lending rates are substitutes. Therefore, for c c c
i
< <
min
, the zero-profit
condition (3) defines a negative linear relation between the collateral-loan ratio
i
c and the
lending rate
i
r . This is plotted as the segment AB in Figure 1. To the left of point A, there is
credit rationing. To the right of point B, the lending rate equals the cost of capital.
All entrepreneurs will borrow, since their participation constraint is always met. To see
this, note that the expected utility level of borrower i is:
[ ] [ ] ) , 1 min( ) 1 ( ) 1 ( ) 1 (
i c i i i i i
c r c p r c p u + + + + + = (6)
[ ] [ ] c c c p r p
c c c r p
i i c i
i i i
< + + +
+ + +

=
if ) 1 ( 1 ) 1 ( ) 1 (
if ) 1 ( ) 1 (

If the individual i does not borrow, however, his utility is just the collateral
i
c . Using

5
Recall the positive-NPV condition r p + > + 1 ) 1 ( . Then, setting 1 = =
p c
in equation (4) implies a
negative
min
c .
6
equations (3) and (6), the participation constraint
i i
c u reduces to 0 ) 1 ( ) 1 ( + + r p .
Given the assumption that NPV > 0, this condition is always met.
Now consider an improvement in judicial efficiency. This can take two forms: an
increase in
c
or in
p
, the fractions of external and internal collateral that lenders can
recover. We examine these two cases in turn.
An increase in
c
shifts the downward-sloping portion of the zero-profit locus inward
from AB to A'B'. The minimum collateral declines to the level corresponding to A', and the
region of credit rationing shrinks: the improvement in judicial efficiency turns some loss-
making loans into viable ones. Borrowers with collateral ratios between
min
c and c already
had access to credit, but now they pay less interest. Therefore, for any given borrower i, the
interest rate either decreases or stays unchanged. However, the average lending rate may also
increase depending on how the composition of the borrowers pool changes as the credit
market expands. The effect on the average rate is negative when initially there is no credit
rationing. This effect is attenuated and can even change sign depending on how many initially
excluded borrowers gain access to credit when increases.
6
Next, consider an increase in
p
. In this case the downward-sloping portion of the zero-
profit locus in Figure 2 expands from AB to A'B. As a consequence, the region of credit
rationing shrinks and lending increases, in contrast with the previous experiment. The rates
charged to those who were already borrowing are unchanged. To understand this difference,
consider that in Figure 1 the increase in
c
implies that borrowers effectively pledge more
external collateral. Since the latter is a substitute for the interest rate, competition forces banks
to lower rates. In Figure 2, instead, borrowers can pledge more internal collateral, which
protects the bank only when the project succeeds. But for borrowers who were not credit-
rationed, banks were already protected by inside collateral in case of success, so the zero-
profit interest rate is unaffected. Borrowers who were previously rationed now have access to
credit at a higher interest rate, since raising the rate is the only way the bank can exploit the

6
To see this, consider two examples. If borrowers collateral-loan ratios are uniformly distributed between
min
c
and c , the average interest rate can be shown to decrease. Suppose instead that there are two groups of potential
borrowers, A and B. Group A is a fraction q of the population and has collateral-loan ratio c c
A
. Group B has
collateral-loan ratio
min
c c
B
< and is drawn into the credit market after the increase in judicial efficiency. It is
immediately clear that in this second example the average interest rate increases from its initial level r .
7
increased inside collateral. Thus, unlike an increase in
c
, an increase in
p
always increases
the average lending rate.
So far we have considered the probability of success as an exogenous parameter p
common to all entrepreneurs: by assumption, judicial efficiency does not affect the default
rate 1p. But in general the probability of a project's success is endogenous, being determined
by entrepreneurial effort to avoid default. Consider a situation where lenders can observe (and
contract upon) the entrepreneurs effort to avoid default, p
i
. In Appendix 1, we show that in
this case judicial efficiency tends to raise the average default rate, although it leaves the
individual default probability unaffected. More specifically, the average default rate increases
whenever there are some entrepreneurs who were denied credit before the judicial reform. The
reason is that a more efficient judiciary reduces the region of credit rationing, opening the
market to lower-grade borrowers.
7
The deterioration of the borrower pool due to this
endogenous response of p
i
tends to raise the average interest rate, acting through a channel
that is absent when p is exogenous. In the case of an increase in
p
this effect reinforces the
increase in the average interest rate. In the case of an increase in
c
, it expands the region of
parameters for which the average interest rate increases.
To summarize, under perfect competition an improvement in judicial efficiency reduces
credit rationing and increases lending. It can also increase the average default rate if there was
prior credit rationing. The effect on interest rates depends on the specifics of the reform: better
recovery of external collateral (
c
) has ambiguous effects, while better recovery of internal
collateral (
p
) raises interest rates.

7
The judicial reform may also raise the default rate via another channel. Banks are more protected by collateral
in case of default, and so have less incentive to screen (collateral and screening being substitutes from their point
of view). Less screening will increase the riskiness of their loans and the average default rate, as shown by
Manove, Padilla and Pagano (2000).
8
2.2 Monopoly
To explore the effects of judicial reform in non-competitive credit markets, consider a
situation in which the credit market is geographically segmented and banks are local
monopolists. Since we assume that the demand for credit is inelastic, the monopolist extracts
from borrower i the entire surplus, setting:
i c i p i
c r + + = + ) 1 ( 1 , for [ ]
max min
, c c c
i
, (7)
where
min
c is given by equation (4). The maximum collateral
max
c that a borrower is willing
to pledge is obtained by substituting equation (7) into the participation constraint:
[ ]
i i c i i i
c c p r c p u + + + + = ) 1 )( 1 ( ) 1 ( ) 1 ( ,
which yields:
c
p
p
c

) 1 )( 1 (
max
+
= . (8)
The interest rate that corresponds to this collateral level is ] ) 1 ( )[ 1 ( 1
max
p p r
p
+ + = + .
Equation (7) shows that, in contrast to the competitive case, under monopoly there is a
positive correlation between the lending rate and the collateral-loan ratio. With no
competition, the bank can charge higher rates to those who pledge more collateral. The
correlation between r
i
and c
i
is graphed as the line AB in Figure 3. As under competition, if
the collateral-loan ratio is lower than
min
c no credit is granted.
Figure 3 illustrates that an increase in
c
shifts the AB locus upward and to the left. The
new locus A'B' features lower
min
c and
max
c . So the credit-rationing region shrinks and
lending increases, as under competition. Lending rates rises for all borrowers. An increase in
c
effectively raises the pledgeable portion of collateral and so enables the bank to extract a
higher surplus by raising interest rates. In Figure 4 we repeat the analysis for an increase in
p
. In this case, the interest rate locus has a parallel upward shift, with similar qualitative
effects. In short, under monopoly an improvement in judicial efficiency reduces credit
rationing, increases lending and raises interest rates.
9
3. Evidence from a Panel of Judicial Districts
The model illustrates that improvements in judicial efficiency reduce credit rationing and
increase aggregate lending. Interest rates can either rise or fall, depending on the competitive
structure of the banking sector and on the specific channel through which judicial reforms
enhance enforcement. We now bring empirical evidence to bear on these issues. In this
section, we use panel data on lending to firms, credit rationing and interest rates in Italian
provinces. In the next section, we will turn to international comparative evidence on mortgage
markets.
3.1. Data
To study the relationship between judicial efficiency and credit market performance, we
merge indicators of efficiency for 27 judicial districts with credit market data for 95 Italian
provinces.
We rely on two indicators of judicial efficiency, using data from the Italian National
Institute of Statistics (ISTAT). The first indicator is the length of ordinary civil trials from
1984 to 1998. It measures the time elapsing between the date of initial recording of a trial and
that of the court sentence, for actions requiring adjudication of substantive rights concerning
credit and commercial matters: loans, sale of real estate or goods, rentals, negotiable and
quasi-negotiable instruments, and insurance.
8
Enforcement cost is directly related to the length
of the judicial process. A long trial increases legal expenses and, for disputed loans, the
interest income forgone when collateral does not cover judicial costs. Moreover, during the
trial, the creditor is exposed to the danger of asset substitution by the debtor and to unexpected
changes in the value of collateral.
The second indicator of judicial efficiency is the number of civil suits pending per
thousand inhabitants. It refers to all actions requiring adjudication of substantive rights,
including appeal trials, from 1984 to 1998. The stock of trials pending is a key determinant of

8
A narrower classification of legal actions (e.g., loans only) produces too few observations for each district-year
cell to compute reliable indicators of judicial efficiency. For the same reason we do not consider the length of
appeals civil cases and bankruptcy procedures.
10
the duration of future trials; the two indicators are strongly correlated.
Our indicators of judicial efficiency vary considerably with judicial district and over
time. The two graphs in Figure 5 display the national averages. The length of trials doubles
from 26.3 months in 1984 to 52.9 in 1998. The number of trials pending per 1000 inhabitants
increases from 23.4 in 1984 to 37.9 in 1996, then edging down to 35.7 in 1998. These trends
may be explained by increasing assignment of judges and resources to criminal justice, to the
increasing number and complexity of civil laws, and to litigation.
The graphs in Figure 6 break down the time series of the two indicators geographically.
Trials are longer and backlogs are larger in the South and the Islands than in the North and
Center. While the difference in the length of trials across regions is roughly constant, the
backlog shows widening geographical disparities. In 1984 the number of trials pending was 20
in the North and 27 in the South; in 1998, 23 in the North and 44 in the South. Furthermore,
the North shows more marked signs of improvement after 1993, when its backlog peaked at
27.4.
The bottom graphs in Figures 5 and 6, then, show that both the average number and the
dispersion of trials pending increased between 1984 and 1998. In a panel regression
framework, variability of the length of trials between different years and different districts is
crucial to identify the effect of judicial efficiency on credit market performance.
Both of our indicators may suffer from measurement error. The cases used to measure
length include many disputes on matters other than credit. The stock of trials pending refers to
the even broader aggregate of all civil cases. Indirect evidence on the reliability of these
indicators comes from a 1994 survey of 269 Italian banks, representing 90 per cent of total
loans.
9
The survey was designed by the Bank of Italy to gather information on credit recovery
costs and procedures (both judicial and non-judicial), in the presence of insolvent borrowers.
It allows us to compare our measures of judicial efficiency, which are based on ISTAT data,
with the bank's own assessment of the length of the judicial procedures by region. Since the
survey refers to 1994, we aggregate the ISTAT district-level judicial data by region (20 in
total, with 1 to 9 provinces each) and relate the resulting measures to the self-reported
indicator.
Figure 7 shows that the length of trials and the stock of trials pending based on ISTAT

9
Generale and Gobbi (1996) describe the survey and its main findings.
11
data correlate positively with the bank's reports. The self-reported measure of the length of
trial has a 0.79 correlation with the ISTAT measure of the same variable (statistically
significant at the 1 percent level) and a 0.45 correlation with the ISTAT-based backlog
(significant at the 5 percent level). We take this as evidence that the two ISTAT-based
indicators of judicial efficiency used in our empirical analysis track lenders perceived credit
collection costs reasonably well.
10
We merge these indicators of judicial efficiency with measures of credit market
performance: outstanding loans, indicators of credit rationing, interest rates on short-term
loans to non-financial companies, ratio of non-performing to total loans and the Herfindhal
index of loan concentration.
Loans granted is total lending to domestic companies in each province divided by
provincial GDP. Credit rationing is proxied by the proportion of overdrawn credit lines to
non-financial firms in each province, that is, lines for which credit is drawn above the amount
initially granted by the bank. This is widely regarded as a good indicator of the tightness of
the credit market because the cost of credit rises steeply when firms overdraw. Interest rates
are provincial averages weighted by loans. The ratio of non-performing loans to total loans is
a proxy of the default rate. All these variables are drawn from the database of the Italian
public credit register (Centrale dei Rischi: see Appendix 2 for details on data sources and
definitions). They are aggregated, for the 95 Italian provinces, from 1984 to 1995.
Table 1 reports unweighted provincial averages of the variables used in the empirical
analysis for three sub-periods. The total number of observations is 1,140 (95 provinces for 12
years). The ratio of total outstanding lending to GDP increases from 31 to 41 percent. Credit
rationing also increased, possibly a reflection of monetary policy tightening during Italys run-
up to the European Monetary Union. Both the lending rate and the T-bill rate decline over the
sample period, reflecting disinflation. The differential between the two also narrows from 5 to
3.6 percentage points. The Herfindhal index declines from 17 to 15 percent, revealing
increased competition in the loan market.

10
The self-reported indicator cannot be directly used in our regression analysis because it is available for only
one year. Therefore, this variable is not identified in a panel data framework.
12
3.2. Descriptive Evidence
Figures 8, 9, 10 and 11 report evidence on the relation between credit market performance and
judicial efficiency in the various judicial districts. Averages are taken over the 1984-95 period.
Figure 8 indicates that the district average amount of lending is negatively correlated with the
length of trials in a district and with the stock of trials pending. The correlation is statistically
different from zero at standard significance levels. For instance, in a relatively efficient
judicial district like Venice where trials last slightly more than 30 months and there are about
22 pending trials per 1000 inhabitants, lending is over 40 percent of GDP. In Reggio Calabria,
where length exceeds 50 months and the backlog is about 50 trials per 1000 inhabitants,
lending is equal to just 10 percent of GDP.
Figure 9 indicates that where trials are longer and the judicial backlog is heavier, our
indicator of credit rationing is also higher: moving from Venice to Reggio Calabria, it
approximately doubles. In Figure 10 we relate the interest rate spread (the difference between
the lending rate and the T-bill rate) to the same indicators of judicial efficiency. Both
correlation coefficients are positive and statistically different from zero at standard
significance levels. The spread is more than 200 basis points greater in the least than in the
most efficient districts. Figure 11 shows that, like the spread, the non-performing loan ratio is
higher where courts are less efficient.
This descriptive evidence suggests that judicial efficiency is associated with a larger
amount of lending, less credit rationing and lower interest rates, in accord with the model of
Section 1 assuming banking competition. However, these relations could be spurious, because
so far we have not controlled for other determinants of credit market performance.
Furthermore, the cross-sectional evidence does not exploit the time-series dimension of the
data set. As we shall see, this feature allows us to control not only for other covariates, but
also for unobserved heterogeneity at the provincial level. Therefore, we turn to regression
analysis.
3.3. Regression Analysis
In our regression analysis we relate lending, fraction of firms with overdraft loans, interest
13
rates and non-performing loans to length of trials and judicial backlog, controlling for credit
market concentration, provincial GDP, calendar-year effects, and in some specifications
provincial effects. Other things equal, we expect market concentration to reduce lending and
raise interest rates, reflecting a less competitive credit market and possibly closer bank-firm
relations, a further channel for higher interest rates and less lending according to Petersen and
Rajan (1995). One would expect a larger GDP to increase the demand for loans and thereby
raise interest rates. To avoid endogeneity, the GDP variable is lagged. Calendar-year dummies
control for the effect of aggregate shocks on the credit market.
The upper panel of Table 2 reports OLS estimates, while the lower panel reports fixed-
effect estimates, which control for unobserved heterogeneity at the province level. In the OLS
regressions, the length of trials and the size of the backlog are associated with less lending,
more overdraft loans, wider spreads and higher default rates. Each of these effects is
statistically different from zero at the 1 percent level, and all are in keeping with the
descriptive evidence of the previous section. In economic terms, moreover, the coefficients are
very large. For instance, increasing the length of trials by 1 year is associated with a fall in the
lending-GDP ratio of almost 7 percentage points and a 2-point increase in the percentage of
firms with overdraft credit. An extra 10 trials pending per 1000 inhabitants is associated with
a reduction of 4 percentage points in the lending ratio and an increase of over 2 in the
overdraft percentage. The positive correlation between efficient judicial districts, lending and
overdraft credit dovetails with the prediction of the model, insofar as overdrafts proxy for
credit rationing.
The Herfindhal index is positively correlated with lending, interest rates and default
rates, and is negatively correlated with the percentage of firms using overdraft credit, although
the coefficient is statistically different from zero only for interest rates. This is consistent with
previous studies of the Italian credit market based on individual loan contract data.
11
In
general, the GDP coefficients are not statistically different from zero.
These results are subject to the objection that judicial efficiency could be correlated with
omitted variables at the provincial level, such as credit risk or the efficiency of banks. The
fixed-effect regressions reported in the lower panel of Table 2 control for such unobserved
heterogeneity, provided that the variation of judicial efficiency in each province over time is

11
See De Bonis and Ferrando (1997), DAuria and Foglia (1997) and Sapienza (1997).
14
not correlated with that of these omitted variables. As one would expect, the impact of judicial
efficiency on all the credit market variables is much attenuated compared with the OLS
estimates and the descriptive evidence of the previous section.
However, the coefficient of the stock of trials pending maintains the same sign and
remains significant in both the lending and the overdraft regressions. An additional 10 trials
pending per 1000 inhabitants is associated with a reduction of 1.5 points in the lending-GDP
ratio and an increase of 1 point in the percentage of firms with overdraft loans. Judicial
backlog correlates negatively with the interest rate spread and with non-performing loans,
overturning the descriptive evidence of Figures 10 and 11 and the OLS estimates, but this is
quite reasonable in the framework of our model. Recall that in the model judicially less
efficient districts may have lower average interest rates under competition, and should have
lower rates under monopoly. Moreover, with the default rate endogenous, the model also
predicts that they will be lower in the less efficient districts.
A caveat is that if the judicial process is excessively long or costly, private parties may
bypass the courts for alternative forms of dispute settlement. The substitution of out-of-court
settlement could be particularly significant in bankruptcies, suggesting that the relation
between credit conditions and judicial enforcement may be non-linear. For short or moderate
trials times, credit market performance (loans, interest rates, and so forth) respond to our
indicators of judicial efficiency. Since beyond a critical length the relation between judicial
efficiency and credit market performance may weaken or disappear, we introduce quadratic
terms in the indicators of judicial efficiency in the specification of Table 2, but these prove to
be not statistically different from zero.
To summarize, the econometric estimates obtained controlling for unobserved
heterogeneity via province-level fixed effects confirm only part of the descriptive evidence of
Figures 7 to 11. According to the estimates, the judicial districts with better legal enforcement
display more lending activity and less credit rationing. These results are consistent with the
model of Section 2, which predicts that judicial efficiency will increase lending and decrease
credit rationing under competition and monopoly alike. On the whole, the estimates suggest
that the correlation of judicial efficiency with interest rates and default rates is less robust.
Again, this is in line with the models predictions.
These results are also consistent with the findings of studies of other countries and
15
markets. Castelar Pineiro and Cabral (2001) and Cristini, Moya and Powell (2001) analyze
how local variations in the effectiveness of the legal system in Argentina and Brazil have
affected the development of credit markets. They find less lending and more non-performing
loans in provinces or states with poor enforcement. Similar results are reported for household
credit in the U.S. and Italy. In the United States, Meador (1982) and Jaffee (1985) found that
mortgage interest rates were generally higher in states where the foreclosure process was
longer and more costly. In Italy, Fabbri and Padula (2001) find that households located in
judicially less efficient districts receive less credit, even after controlling for household
characteristics.
4. International Evidence on Mortgage Markets
The market for mortgage loans is a potentially fruitful testing ground for the effects of the
quality of judicial enforcement. First, this market is relatively homogeneous internationally, so
that comparison is meaningful. Second, in the mortgage market an indicator of credit rationing
is readily available: namely, the minimum down payment ratio. Finally, the performance of
mortgage markets can be related to a set of indicators of the effectiveness of foreclosure
procedures, available for a good number of industrialized countries.
The first three columns of Table 3 report the ratio of outstanding mortgage lending to
GDP, the down payment ratio, and the spread between the lending and the borrowing rate in
fourteen countries. Mortgage markets differ widely. In Canada, the United States, the United
Kingdom, Sweden and Finland, the market is well developed, and the down payment is
relatively low. In other countries (Belgium, Italy, Germany and Spain) the market is relatively
thin and the down payment ratio is high.
The spread between borrowing and lending rates is an important indicator of mortgage
market imperfections. Per se, a spread is not inconsistent with equilibrium models: it can stem
from transaction costs or imperfect competition, and is negatively correlated with the
equilibrium supply of loans. However, the presence of a spread is also consistent with
asymmetric information or opportunistic behavior by borrowers. In these alternative models,
there is no necessary relation between the spread and the supply of loans. For instance, in the
model developed in Section 2 the spread is due to a particular form of transaction cost
16
(judicial inefficiency), but it can widen as well as narrow following judicial improvement.
In our sample of fourteen countries, differences in spreads are relatively small, while the
variation in mortgage lending is huge. The spread ranges only from 2.3 percentage points in
Spain (a country with comparatively low mortgage debt) to about 1.5 points in Italy and
Austria (also with low levels of debt) and the United States (which has the largest mortgage
debt). In short, there is simply no correlation between the spread and the size of the mortgage
market. By contrast, the down payment ratio exhibits considerable variability. It is highest in
Austria, Belgium, Germany, Italy, and Luxembourg, which unsurprisingly also have
comparatively small mortgage markets.
12
The lack of correlation between the spread and the
volume of lending (coefficient of 0.04) and the strong, negative correlation of the down
payment ratio to volume (0.63) are consistent with the model of Section 2.
Cross-country variability in the volume of mortgage lending, down payment ratios and
interest rate spreads can be traced both to supply factors, including the cost and speed of
foreclosure procedures, and to demand factors and regulation. Key demand factors are
household earnings profiles, the age structure of the population, ownership preference, tax
incentives for homeownership and debt, and intergenerational transfers. Regulation often
imposes interest rate controls and minimum down payment ratios (until 1986 it was 50
percent in Italy). Here the analysis is descriptive, and we focus on international differences in
judicial enforcement to explain the different performance of mortgage markets, without
controlling for these additional factors.
Section 2 emphasizes that differences in the cost of repossessing and liquidating
collateral can affect the performance of credit markets. Table 4 reports three indicators of
judicial efficiency in the various countries. The first is a survey-based general assessment of
the quality of judicial enforcement. The others are the length and the average cost of
foreclosures on home mortgage loans.
On the basis of these indicators, Belgium, Germany, Italy and Spain feature more costly

12
Chiuri and Jappelli (2001) explore the determinants of the international pattern of home ownership using the
Luxembourg Income Study (LIS), a set of microeconomic data which they merge with aggregate panel data on
mortgage loans and down payment ratios for fourteen OECD countries. After controlling for demographic
characteristics, country effects, cohort effects and time effects, they find strong evidence that the availability of
mortgage finance as measured by outstanding mortgage lending and down payment ratios affects the age-
profile of home ownership, especially at the young end.
17
and slower procedures, and less efficient judicial systems in general.
13
The Italian case stands
out. Consistent with the data reported in Section 3, debt collection and repossession in case of
foreclosure is very costly and time-consuming. It takes between 3 and 5 years to repossess,
and legal expenses for foreclosure can be as high as 20 percent of the price. At the other
extreme, the Netherlands, Canada, the United States and the United Kingdom have rapid
mortgage foreclosure (one year or less, with a minimum of 2-3 months in the Netherlands)
and much cheaper procedures.
Figure 12 plots the ratio of mortgage lending to GDP against two of our three indicators
(duration of foreclosure and overall judicial efficiency). The size of the mortgage market
correlates negatively with duration and positively with judicial efficiency: that is, the countries
with better judicial systems also feature the broadest mortgage markets. Figure 13 suggests
that judicial efficiency correlates negatively with down payment ratios, duration positively.
That is, the countries with better judicial systems also have less credit rationing. Figure 14
shows that the spread correlates negatively with duration and positively with efficiency. That
is, in countries with better judicial systems interest rates on loans are relatively higher.
14
The
patterns of Figures 12, 13 and 14 are summarized by the correlation matrix reported in Table
4. The correlations of lending volume and down payments with the three judicial efficiency
indicators are statistically different from zero at standard significance levels. For spreads, only
the positive correlation with the overall index of judicial efficiency is statistically different
from zero.
The descriptive evidence reported in this section suggests that enforcement problems
may be at the roots of the international differences in mortgage lending and in downpayment
ratios. The evidence is consistent with the predictions of the theoretical model and with our
findings for the panel of Italian provinces. It is also consistent with the findings of other recent

13
The three indicators of judicial efficiency are strongly correlated. For instance, duration correlates negatively
with efficiency and positively with legal expense (see Table 4).
14
In contrast to the international comparison, some studies of mortgage markets in the United States report
evidence that the cost of legal enforcement increases the cost of credit. Meador (1982) and Jaffee (1985) find
that mortgage interest rates were generally higher in states where the law extended the length and expense of the
foreclosure process. Similarly, Gropp, Scholz and White (1997) document that in states with more generous
bankruptcy exemptions low-wealth households receive less credit and are charged higher interest rates. Alston
(1984) reports that farm foreclosure moratorium laws in the 1930s led to fewer farm loans and to higher interest
rates in the states that enacted them. Consistently with these findings, in states that facilitate the foreclosure
process the rate of foreclosure is higher (Clauretie, 1987) and the losses incurred by lenders are lower (Clauretie
and Herzog, 1990).
18
papers based on cross-country data. La Porta et al. (1997) consider indicators of creditor
protection, origin of the legal system and respect of the law to explain the private debt-GNP
ratio, using cross-country data, and find that respect of the law has a large and statistically
significant effect on the size of the capital market (p. 1145). Padilla and Requejo (2001)
produce estimates that qualify those results, using the same basic data but also controlling for
variables that capture macroeconomic stability. In these expanded specifications, the
efficiency of judicial enforcement appears to have more significant effects on credit markets
than creditor protection per se, in contrast with the original La Porta et al. (1997) results.
5. Conclusions
Judicial inefficiency has high economic costs in credit markets. So far, these costs have never
been measured. This paper takes a step in this direction by analyzing the effect of judicial
efficiency on the availability and cost of credit, using a model of opportunistic debtors and
inefficient courts. The model illustrates that improvements in judicial efficiency reduce credit
rationing and increase the volume of lending. Interest rates can either increase or decrease,
depending on the competitive structure of banks, on the specific channel through which
judicial reforms improve lenders ability to repossess collateral, and on composition effects.
For instance, greater judicial efficiency can open up the credit market to low-grade borrowers
previously judged not creditworthy, and thereby raise the average default rate and the average
interest rate.
These theoretical predictions receive support from panel data on Italian provinces and
international data on mortgage markets. We construct a panel of Italian provinces merging
judicial and credit market data. Controlling for unobserved heterogeneity at the provincial
level, we find that where the backlog of pending trials is relatively large credit is less widely
available, while the average interest rate and the default rate are lower.
International data also reveal that the depth of mortgage markets and the availability of
mortgage credit related inversely to the costs of foreclosure and directly to indicators of
judicial efficiency, providing further evidence that judicial efficiency is associated with
financial market deepening and more abundant credit.
19
Appendix 1: The Model with Endogenous Default
Assume that the utility of entrepreneur i is:
[ ] [ ] ) ( ) , 1 min( ) 1 ( ) 1 ( ) 1 (
i
p
i
V
i
c
c i
r
i
c
i
p
i
r
i
c
i
p
i
u + + + + + = (A1)
where the disutility of effort V
i
(p
i
) is an entrepreneur-specific, increasing and convex function of the
success rate p
i
. We assume that p
i
and c
i
are observable and that the terms of the contract can be
conditioned upon them. Therefore the competitive interest rate charged to entrepreneur i reflects both.
Entrepreneur i chooses his effort level p
i
treating this interest rate r
i
as an exogenous parameter. The
first-order condition of the problem is:
[ ] [ ] 0 ) ( ) , 1 min( ) 1 ( ) 1 (
'
= + + + + =

i i i c i i i i
i
i
p V c r c r c
p
u
(A2)
The second-order condition for a maximum is satisfied due to the convexity of V
i
(p
i
).
The competitive interest rate is given by:
) , 1 min(
1 1
1
i c i
i
i
i
i
c r
p
p
p
r
r +

+
= + for
i i
c c
min,
, (A3)
where
c
p i
c
i
p
r
c

) 1 (
1
min,
+

+
= (A4)
is the minimum collateral that entrepreneur i must pledge to obtain credit. The higher the effort p
i
, the
lower the minimum collateral. In contrast with the case with constant p analyzed in the text (where the
marginal borrower is identified only by his collateral), here condition (A4) identifies a set of marginal
borrowers. All entrepreneurs with collateral c
i
and success rate p
i
that satisfy equation (A4) are
marginal borrowers.
Replacing the competitive interest rate (A3) in the first-order condition (A2), one obtains the
equilibrium success rate of any entrepreneur i:
+ =1 ) (
'
i i
p V (A5)
irrespective of whether
i c
c is smaller or larger than
i
r + 1 . Condition (A5) establishes that, at the
individual level, the equilibrium success rate depends only on project profitability and on preferences,
and not on judicial efficiency. However, an increase in judicial efficiency can affect the average
success rate via composition effects, depending on the prevalence of credit rationing prior to the
20
reform. From condition (A4), an increase in
c
or in
p
reduces the minimum required collateral c
min,i
(given p
i
) or, alternatively, reduces the minimum required effort p
i
(given c
min,i
). Thus, a new group of
borrowers will gain access to credit: they feature lower c
i
, lower p
i
or both. It follows that the average
default rate of the pool of borrowers increases, whenever some borrowers were credit-rationed before
the judicial reform. If, instead, no entrepreneurs were credit-rationed (c
i
> c
min,i
for all i), then the
average default rate remains unchanged.
The interest rate charged to each individual borrower i rises along with his default rate. To see
this, notice that the interest rate charged to entrepreneur i is a decreasing function of his probability of
success p
i
, and therefore an increasing function of his default rate:

< <

=
+ +
<
+

+
. if 0
) ( ) 1 (

, if 0
1
) 1 (
2 2
2
c c
p
c c
p
c r
c c
p
r
p
r
i
i
i c
i
i c
i
i
i
i

21
Appendix 2: Provincial Data
Credit market data are available for 95 Italian provinces for the period 1984-95. The data are drawn
from the Centrale dei Rischi database. The Centrale dei Rischi is the Italian central credit register,
managed by a department of the Bank of Italy. Between 1984 and 1995 it recorded data on each loan
over 80 million lire (approximately Euro 40,000) granted by Italian banks to companies and
individuals. These data are compulsorily filed by banks and made available upon request to individual
banks to monitor the total exposure of their customers. In addition, 88 banks (accounting for over 70
percent of total bank lending) have agreed to file detailed information about the interest rates charged
on each loan. These data, which are collected for monitoring purposes, are highly confidential.
Judicial data are available from 1984 to 1998 for 27 judicial districts. Each district is defined by the
jurisdiction of an appeal's court and comprises one or more provinces. Table 5 reports the matching of
provinces and judicial districts. Below we report the definition and source of the variables used in the
estimation.
Length of trials, by judicial district (1984-98). Interval between the date of initial filing of a civil
action and the date of the sentence, for actions requiring adjudication of substantive rights concerning
the following matters: loans, sale of real estate or goods, rentals, negotiable and quasi-negotiable
instruments, and insurance. Source: data kindly provided by the Italian National Institute of Statistics
(ISTAT).
Stock of pending trials, by judicial district (1985-98). Number of pending civil trials, based on actions
requiring adjudication of substantive rights and scaled by the population of the corresponding district.
Source: Annuario Statistico dei Procedimenti Giudiziari Civili, various years, Italian National
Institute of Statistics (ISTAT).
Loans granted, by province (1984-95). Total credit granted to domestic companies for loans above 80
millions lire. Source: Centrale dei Rischi.
Credit rationing, by province (1985-95). Proportion of credit lines overdraft (loans for which credit
actually drawn exceeds credit granted) for a set of non-financial companies. The companies are those
that are also present in the Company Account Data Service Centrale dei Bilanci, covering
approximately 30,000 companies each year. Source: Centrale dei Rischi.
Lending rate, by province (1984-95). Lending rate on short-term loans in domestic currency to
domestic companies, for a sample of 88 banks that reports on quarterly lending rates on loans
exceeding 80 million lire. Data are aggregated by province weighting interest rates by loan size.
Annual data are computed as averages of quarterly data. Source: Centrale dei Rischi.
Non-performing loans, by province (1984-95). Ratio of non-performing loans to total loans in
domestic currency to domestic companies. Annual data are computed as averages of quarterly data.
Source: Centrale dei Rischi.
Herfindhal index, by province (1985-95). The index is the sum of squared market shares of loans of
all banks in each province. Source: Centrale dei Rischi.
Real GDP, by province (1985-95). Source: Banca dItalia estimates based on data from Istituto
Tagliacarne. The estimation method is described by Fabiani and Pellegrini (1997).
22
References
Alston, Lee (1984), Farm Foreclosure Moratorium Legislation: A Lesson from the Past,
American Economic Review 74, 445-457.
Castelar Pineiro, Armando and Celia Cabral (2001), Credit Markets in Brazil: The Role of
Judicial Enforcement and Other Institutions, in Marco Pagano (ed.), Defusing Default:
Incentives and Institutions. Washington: Johns Hopkins University Press, 2001.
Chiuri, Maria Concetta and Tullio Jappelli (2001), Financial Market Imperfections and
Home Ownership: A Comparative Analysis, CEPR Discussion Paper no. 2717.
Clauretie, Terrence M. (1987), The Impact of Interstate Foreclosure Cost Differences and the
Value of Mortgages on Default Rates, Journal of the American Real Estate and Urban
Economics Association 15, 152-67.
Clauretie, Terrence M. and Thomas Herzog (1990), The Effect of State Foreclosure Laws on
Loan Losses: Evidence from the Mortgage Insurance Industry, Journal of Money,
Credit and Banking 22, 221-233.
Cristini, Marcela, Ramiro A. Moya and Andrew Powell (2001), The Importance of an
Effective Legal System for Credit Markets: the Case of Argentina, in Marco Pagano
(ed.), Defusing Default: Incentives and Institutions. Washington: Johns Hopkins
University Press, 2001.
DAuria, C. and A. Foglia (1997), Le determinanti del tasso di interesse sui crediti alle
imprese, in Ignazio Angeloni et al. (eds.), Le banche e il finanziamento delle imprese.
Bologna: Il Mulino.
De Bonis, R. and A. Ferrando (1997), Da che cosa dipendono i tassi di interesse sui prestiti
nelle provincie? Bank of Italy: Temi di Discussione n. 319.
European Mortgage Federation (1996), Comparative Study on Real Estate Enforcement
Procedure in the EC Countries. Brussels: EC Mortgage Federation.
European Mortgage Federation (1997), Hypostat 1986-1996. Brussels: EC Mortgage
Federation.
Fabiani, S., and G. Pellegrini (1997), Education, Infrastructure, Geography and Growth: An
Empirical Analysis of the Development of Italian Provinces, Bank of Italy: Temi di
Discussione n. 323.
Fabbri, Daniela and Mario Padula (2001), Judicial Costs and Household Debt, University of
Salerno, CSEF Working Paper, July.
Generale, Andrea, and Giorgio Gobbi (1996), Il recupero dei crediti: costi, tempi e
23
comportamenti delle banche, Bank of Italy: Temi di Discussione n. 265.
Gropp, Reint, John Karl Scholz and Michelle J. White (1997), Personal Bankruptcy and
Credit Supply and Demand, Quarterly Journal of Economics, 112(1), 217-251.
Jaffee, Austin (1985), Mortgage Foreclosure Law and Regional Disparities in Mortgage
Financing Costs, Pennsylvania State University, Working Paper n. 85-80.
Jappelli, Tullio and Marco Pagano (1994), Saving, Growth and Liquidity Constraints, The
Quarterly Journal of Economics 109, 83-109.
La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer and Robert W. Vishny (1997),
Legal Determinants of External Finance, Journal of Finance 52, 1131-50.
La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer and Robert W. Vishny (1998),
Law and Finance, Journal of Political Economy 106, 1113-55.
Maclennan D., John Muellbauer and M. Stephens (1999), Asymmetries in Housing and
Financial Market Institutions and EMU, CEPR Discussion Paper n. 2062.
Manove, Michael, Jorge Padilla, and Marco Pagano (2001), Collateral vs. Project Screening:
A Model of Lazy Banks, RAND Journal of Economics, Vol. 32, No. 4, Winter, 726-
744.
Meador, Mark (1982), The Effect of Mortgage Laws on Home Mortgage Rates, Journal of
Economics and Business 34, 143-148.
Petersen, Mitchell A., and Raghuram G. Rajan (1995), The Effect of Credit Market
Competition on Lending Relationships, The Quarterly Journal of Economics 110, 407-
43.
Sapienza, Paola (1997), Le scelte di finanziamento delle imprese italiane, in Ignazio
Angeloni et al. (eds.), Le banche e il finanziamento delle imprese. Bologna: Il Mulino.
24
Table 1
Panel of Italian Provinces: Descriptive Statistics
The table reports unweighted period averages of the variables used in the regression analysis. See Appendix 2 for
the definition of the variables.
Variable 1984-87 1988-91 1992-95
Length of trials, months 30.00 40.08 44.15
Stock of pending trials, per thousand inhabitants 23.55 29.61 34.98
Loans granted/GPD, percent 31.23 39.75 40.67
Credit overdrafts, percent 11.48 15.23 19.44
Lending rate, percent 17.79 15.42 14.97
T-bill rate, percent 12.80 12.50 11.23
Non-performing loans/GDP, percent 2.34 1.24 2.14
Herfindhal index, percent 17.33 15.59 15.29
Real GDP (trillion of lire) 11.34 12.54 12.61
Number of observations 380 380 380
25
Table 2
Panel of Italian Provinces: Regression Analysis
The dependent variables are the ratio of loans to GDP, an indicator of credit rationing (the fraction of loans for
which credit used exceeds 100 percent of credit granted), the spread between the lending rate and the T-bill rate,
and the ratio of values of non-performing loans to total loans. All variables are in percent. Each regression is
estimated with a full set of year dummies. The sample consists of observations for 95 provinces from 1984 to
1995. T-statistics are reported in parenthesis.
Ordinary Least Squares Estimates
Variable Lending / GDP Overdrafts Interest rate
spread
Non-performing
loans / total loans
Length of trials, months -0.583
(-7.59)
0.181
(7.57)
0.049
(12.10))
0.047
(5.66)
Stock of pending trials, per
thousand inhabitants
-0.438
(-7.40)
0.214
(11.61)
0.045
(14.53)
0.022
(3.36)
Herfindhal index 0.080
(1.04)
-0.007
(-0.30)
0.014
(3.58)
0.010
(1.21)
First lag of real GDP 2.400
(3.18)
-0.072
(-0.31)
-0.022
(-0.55)
0.029
(0.35)
Second lag of real GDP -0.756
(-0.98)
0.038
(0.16)
0.001
(0.03)
-0.037
(-0.44)
Adjusted R square 0.760 0.519 0.676 0.167
Fixed Effect Estimates
Variable Lending / GDP Overdrafts Interest rate
spread
Non-performing
loans / total loans
Length of trials, months -0.002
(-0.05)
0.011
(0.40)
0.007
(1.90)
-0.012
(-0.98)
Stock of pending trials, per
thousand inhabitants
-0.147
(-2.86)
0.106
(3.72)
-0.005
(-1.47)
-0.045
(-3.45)
Herfindhal index -0.209
(-3.11)
0.113
(3.01)
0.001
(0.25)
0.054
(3.14)
First lag of real GDP -0.451
(-1.46)
-0.118
(-0.69)
0.026
(1.16)
0.011
(0.14)
Second lag of real GDP -0.238
(-0.81)
-0.055
(-0.34)
0.001
(0.01)
-0.069
(-0.92)
26
Table 3
Housing Finance, Costs and Duration of Housing Mortgage Foreclosure, and Efficiency
of the Judicial System: International Comparison
Outstanding mortgage loans over GDP are 1986-96 averages. Annual outstanding loans against mortgages in
residential property is based on Table 14 in EU Mortgage Federation - Hypostat 1986-96 (1997) and annual GDP
from IMF Financial Statistics. The downpayment ratio is the 1970-1995 average of minimum downpayment
ratios for first-time buyers. The sources are Jappelli and Pagano (1994), EC Mortgage Federation (1996) and
Maclennan, Muellbauer and Stephens (1998). Data refer to 1981-97. The interest rate spread is the average
interest rate on mortgage loans less the corresponding long-term rate. Interest rates on mortgage loans are drawn
from Hypostat 1986-96, Table 21. Long-term interest rates are drawn from OECD (1996). Data refer to 1986-96,
except for Finland and Sweden (1990-96), Luxembourg (1986-87) and Spain (1993-96). Efficiency of the
judicial system is an assessment of the entire legal environment as it affects business taken from the country-risk
agency Business International Corporation. It is an average of 1980-83 and the scale goes from 0 to 10, with
lower scores indicating lower efficiency levels. Source: La Porta et al. (1998). Legal expenses as percentage of
the price of the mortgaged house and duration of housing mortgage foreclosure refer to 1990 and are drawn from
European Mortgage Federation (1996). Data for duration in Austria, Canada, Luxembourg, and United States
have been obtained directly by country experts.
Country
Outstanding
mortgage
loans / GDP
Down-
payment
ratio
Interest rate
spread on
mortgage
loans
Efficiency
of the
judicial
system
Duration of
mortgage
foreclosure
(in months)
Legal
expenses as
% of the
mortgaged
house price
Australia 19.30 20.0 n.a. 10 n.a. n.a.
Austria 4.24 30.0 1.52 9.5 13 n.a.
Belgium 20.08 22.5 1.02 9.5 24 16-23
Canada 41.32 22.5 n.a. 9.25 4.75 n.a.
Finland 32.35 17.5 1.23 10 n.a. n.a.
France 22.02 20 0.95 8 10-12 12-18
Germany 28.92 27.5 1.10 9 12-18 6
Italy 5.49 42 1.47 6.75 36-60 18-20
Luxembourg 25.61 40 -1.02 n.a. 12 2
Netherlands 43.29 25 0.41 10 2-3 11
Spain 15.01 20 -2.30 6.25 36 5-15
Sweden 56.50 15 0.20 10 n.a. n.a.
United Kingdom 51.87 9 1.08 10 12 4.75
United States 43.61 15.5 1.60 10 9 n.a.
27
Table 4
International Comparison of Mortgage Markets: Correlation Matrix
The table reports the correlation matrix between indicators of housing finance (mortgage loans, downpayment
ratios and interest rate spreads) and indicators of judicial efficiency (efficiency of judicial system, duration of
mortgage foreclosure, and legal expenses as a percent of the mortgaged house price). The countries analyzed are
the 14 countries listed in Table 3. Because of missing data, some of the correlation coefficients are obtained with
fewer observations. The number in parenthesis is the significance level of each correlation coefficient.
Outstanding
mortgage
loans / GDP
Down-
payment
ratio
Interest rate
spread on
mortgage
loans
Efficiency
of the
judicial
system
Duration of
mortgage
foreclosure
(in months)
Legal
expenses as
% of the
mortgaged
house price
Outstanding
mortgage loans /
GDP
1.0000
Down-payment ratio -0.6310
(0.0150)
1.0000
Interest rate spread
on mortgage loans
0.0482
(0.8818)
-0.0768
(0.8126)
1.0000
Efficiency of the
judicial system
0.5969
(0.0313)
-0.4998
(0.0820)
0.5159
(0.1043)
1.0000
Duration of mortgage
foreclosure (months)
-0.6737
(0.0230)
0.3944
(0.2300)
-0.1977
(0.5841)
-0.8105
(0.0045)
1.0000
Legal expenses as %
of the mortgaged
house price
-0.5694
(0.1407)
0.1015
(0.8110)
0.3953
(0.3324)
-0.3016
(0.5110)
0.5071
(0.1996)
1.0000
28
Table 5
Matching Judicial Districts and Provinces
The table reports the matching of judicial districts with the Italian provinces. The source is ISTAT, Annuario dei
Procedimenti Giudiziari Civili, 1996.
Judicial districts Corresponding regions and provinces Population in judicial districts in 1994
Turin Piedmont (all provinces), Valle dAosta 4,417,412
Milan Milan, Como, Varese, Pavia, Sondrio 6,196,412
Brescia Brescia, Bergamo ,Cremona, Mantua 2,704,486
Trento Trentino-Alto Adige (all provinces) 906,387
Venice Veneto (all provinces) 4,418,139
Trieste Friuli-Venezia Giulia (all provinces) 1,860,380
Genoa Liguria (all provinces) and Massa-Carrara 1,191,768
Bologna Emilia Romagna (all provinces) 3,922,564
Florence Tuscany (all provinces excluding Massa Carrara) 3,326,434
Perugia Umbria (all provinces) 820,529
Ancona Marche (all provinces) 1,440,435
Rome Lazio (all provinces) 5,189,728
LAquila Abruzzo (all provinces) 1,262,802
Campobasso Molise (all provinces) 331,776
Naples Naples, Avellino, Benevento, Caserta 4,633,197
Salerno Salerno 1,080,545
Bari Bari, Foggia 2,248,896
Lecce Lecce, Brindisi, Taranto 1,820,197
Potenza Basilicata (all provinces) 610,082
Catanzaro Catanzaro, Cosenza 1,500,461
Reggio di Cal. Reggio Calabria 578,837
Palermo Palermo, Agrigento, Trapani 2,147,955
Messina Messina 665,591
Caltanisetta Caltanissetta, Enna 443,664
Catania Catania, Ragusa, Siracusa, 1,793,745
Cagliari Cagliari, Oristano 1,068,333
Sassari Sassari, Nuoro 589,765
All districts 57,170,57
29
Figure 1
An increase in recoverable outside collateral (
c
) under competition

A
A'
B
min
c
p
r + 1
i
r + 1
B'
c
i
c
r + 1
30
A
A'
B
Figure 2
An increase in recoverable inside collateral (
p
) under competition

min
c
p
r + 1
i
r + 1
c
i
c
r + 1
31
Figure 3
An increase in recoverable outside collateral (
c
) under monopoly

A
A'
B
min
c
i
r + 1
B'
i
c
max
c
) 1 ( +
p
32
Figure 4
An increase in recoverable inside collateral (
p
) under monopoly

A
A'
B
min
c
i
r + 1
B'
i
c
max
c
) 1 ( +
p
33
Figure 5
Indicators of Judicial Efficiency
The graphs display the average length of ordinary civil trials (in months) and the stock of pending
civil trials (divided by the population of the district) in Italy from 1984 to 1998.

Length of trials: Italy
1985 1990 1995 2000
20
30
40
50
60

Trials pending per thousand inhabitants: Italy
1985 1990 1995 2000
20
25
30
35
40
34
Figure 6
Indicators of Judicial Efficiency, by Region
The graphs display the average length of ordinary civil trials (in months) and the stock of pending
civil trials (divided by the population of the district) in four Italian regions from 1984 to 1998.
Trials pending per thousand inhabitants
North Center
South Islands
1985 1990 1995 2000
20
30
40
50
Length of trials
North Center
South Islands
1985 1990 1995 2000
20
30
40
50
60
35
Figure 7
Comparison between ISTAT and Banks' Reported Measures of Judicial Efficiency
The figures plot the self-reported length of trials against the ISTAT measure of the length of trial and
of the stock of pending trials, respectively. The self-reported length of trial is drawn from a 1994
survey on 269 Italian banks, representing 90 per cent of total lending. Data are grouped by regions (20
in total).
B
a
n
k
s
'

r
e
p
o
r
t
e
d

l
e
n
g
t
h

o
f

t
r
i
a
l
Length of trial: official estimate

30 40 50 60
10
20
30
40
50
Abruzzi
Aosta
Basilica
Calabria
Campania
Emilia
Friuli
Lazio
Liguria
Lombardi
Marche
Molise
Piedmont
Puglia
Sardinia
Sicily
Tuscany
Trentino
Umbria
Veneto
B
a
n
k
s
'

r
e
p
o
r
t
e
d

l
e
n
g
t
h

o
f

t
r
i
a
l
Trials pending: official estimate

20 30 40 50 60
10
20
30
40
50
Abruzzi
Aosta
Basilica
Calabria
Campania
Emilia
Friuli
Lazio
Liguria
Lombardi
Marche
Molise
Piedmont
Puglia
Sardinia
Sicily
Tuscany
Trentino
Umbria
Veneto
36
Figure 8
Judicial Efficiency and Lending
C
r
e
d
i
t

g
r
a
n
t
e
d
Length of trials

30 35 40 45 50 55
0
20
40
60
80
Ancona
Bari
Bologna
Brescia
Cagliari
Caltanissetta
Campobasso Catania
Catanzaro
Florence
Genoa
L'Aquila
Lecce
Messina
Milan
Naples
Palermo
Perugia
Potenza
Reggio Calabria
Rome
Salerno
Sassari
Trento
Trieste
Turin
Venice
C
r
e
d
i
t

g
r
a
n
t
e
d
Trials pending

10 15 20 25 30 35 40 45 50 55 60
0
20
40
60
80
Ancona
Bari
Bologna
Brescia
Cagliari
Caltanissetta
Campobasso Catania
Catanzaro
Florence
Genoa
L'Aquila
Lecce
Messina
Milan
Naples
Palermo
Perugia
Potenza
Reggio Calabria
Rome
Salerno
Sassari
Trento
Trieste
Turin
Venice
37
Figure 9
Judicial Efficiency and Overdraft Loans
P
e
r
c
e
n
t
a
g
e

o
v
e
r
d
r
a
f
t
Length of trials

30 35 40 45 50 55
10
15
20
25
Ancona
Bari
Bologna
Brescia
Cagliari
Caltanissetta
Campobasso
Catania
Catanzaro
Florence
Genoa
L'Aquila
Lecce
Messina
Milan
Naples
Palermo
Perugia
Potenza
Reggio Calabria
Rome Salerno
Sassari
Trento
Trieste
Turin
Venice
P
e
r
c
e
n
t
a
g
e

o
v
e
r
d
r
a
f
t
Trials pending

10 15 20 25 30 35 40 45 50 55 60
10
15
20
25
Ancona
Bari
Bologna
Brescia
Cagliari
Caltanissetta
Campobasso
Catania
Catanzaro
Florence
Genoa
L'Aquila
Lecce
Messina
Milan
Naples
Palermo
Perugia
Potenza
Reggio Calabria
Rome Salerno
Sassari
Trento
Trieste
Turin
Venice
38
Figure 10
Judicial Efficiency and Interest Rate Spread
I
n
t
e
r
e
s
t

r
a
t
e

s
p
r
e
a
d
Length of trials

30 35 40 45 50 55
2
3
4
5
6
Ancona
Bari
Bologna
Brescia
Cagliari
Caltanissetta
Campobasso
Catania
Catanzaro
Florence
Genoa
L'Aquila
Lecce
Messina
Milan
Naples
Palermo
Perugia
Potenza
Reggio Calabria
Rome
Salerno
Sassari
Trento
Trieste
Turin
Venice
I
n
t
e
r
e
s
t

r
a
t
e

s
p
r
e
a
d
Trials pending

10 15 20 25 30 35 40 45 50 55 60
2
3
4
5
6
Ancona
Bari
Bologna
Brescia
Cagliari
Caltanissetta
Campobasso
Catania
Catanzaro
Florence
Genoa
L'Aquila
Lecce
Messina
Milan
Naples
Palermo
Perugia
Potenza
Reggio Calabria
Rome
Salerno
Sassari
Trento
Trieste
Turin
Venice
39
Figure 11
Judicial Efficiency and Non-Performing Loans
N
o
n
-
p
e
r
f
o
r
m
i
n
g

l
o
a
n
s
Length of trials

30 35 40 45 50 55
0
2
4
6
Ancona
Bari
Bologna
Brescia
Cagliari
Caltanissetta
Campobasso
Catania
Catanzaro
Florence
Genoa
L'Aquila
Lecce
Messina
Milan
Naples
Palermo
Perugia
Potenza
Reggio Calabria
Rome
Salerno
Sassari
Trento Trieste
Turin
Venice
N
o
n
-
p
e
r
f
o
r
m
i
n
g

l
o
a
n
s
Trials pending

10 15 20 25 30 35 40 45 50 55 60
0
2
4
6
Ancona
Bari
Bologna
Brescia
Cagliari
Caltanissetta
Campobasso
Catania
Catanzaro
Florence
Genoa
L'Aquila
Lecce
Messina
Milan
Naples
Palermo
Perugia
Potenza
Reggio Calabria
Rome
Salerno
Sassari
Trento Trieste
Turin
Venice
40
Figure 12
Duration of Foreclosure Proceedings, Judicial Efficiency and Mortgage Lending:
International Evidence
M
o
r
t
g
a
g
e

l
o
a
n
s

/

G
D
P
Duration of foreclosure proceedings

0 12 24 36 48
0
10
20
30
40
50
60
Austria
Belgium
Canada
France
Germany
Italy
Luxembourg
Netherlands
Spain
UK
US
M
o
r
t
g
a
g
e

l
o
a
n
s

/
G
D
P
Judicial efficiency

6 7 8 9 10
0
10
20
30
40
50
60
Australia
Austria
Belgium
Canada
Finland
France
Germany
Italy
Netherland
Spain
Sweden
UK
US
41
Figure 13
Duration of Foreclosure Proceedings, Judicial Efficiency and Downpayment Ratios:
International Evidence
D
o
w
n
p
a
y
m
e
n
t

r
a
t
i
o
Duration of foreclosure proceedings

0 12 24 36 48
10
20
30
40
50
Austria
Belgium Canada
France
Germany
Italy
Luxembourg
Netherlands
Spain
UK
US
D
o
w
n
p
a
y
m
e
n
t

r
a
t
i
o
Judicial efficiency

6 7 8 9 10
10
20
30
40
50
Australia
Austria
Belgium Canada
Finland
France
Germany
Italy
Netherland
Spain
Sweden
UK
US
42
Figure 14
Duration of Foreclosure Proceedings, Judicial Efficiency and Interest Rate Spread:
International Evidence
I
n
t
e
r
e
s
t

r
a
t
e

s
p
r
e
a
d
Duration of foreclosure proceedings

0 12 24 36 48 60
-2
-1
0
1
2
Netherlands
US
France
UK
Luxembourg
Austria
Germany
Belgium
Spain
Italy
I
n
t
e
r
e
s
t

r
a
t
e

s
p
r
e
a
d
Judicial efficiency

6 7 8 9 10
-2
-1
0
1
2
Austria
Belgium
Finland
France
Germany
Italy
Netherland
Spain
Sweden
UK
US