Volume 12 Issue 1 Article 2 
12-31-2010 
Financing constraints, credit rationing and financing obstacles: 
evidence from firm-level data in South-Eastern Europe 
Iraj Hashi 
Valentin Z. Toçi 
Follow this and additional works at: https://www.ebrjournal.net/home 
Recommended Citation 
Hashi, I., & Z. Toçi, V. (2010). Financing constraints, credit rationing and financing obstacles: evidence 
from firm-level data in South-Eastern Europe. Economic and Business Review, 12(1). https://doi.org/
10.15458/2335-4216.1241 
This Original Article is brought to you for free and open access by Economic and Business Review. It has been 
accepted for inclusion in Economic and Business Review by an authorized editor of Economic and Business 
Review. 
       ECONOMIC AND BUSINESS REVIEW  |  VOL. 12  |  No. 1  |  2010  |  29–60 29
FINANCING CONSTRAINTS, CREDIT RA-
TIONING AND FINANCING OBSTACLES: 
EVIDENCE FROM FIRM-LEVEL DATA IN 
SOUTH-EASTERN EUROPE
IRAJ HASHI*
VALENTIN Z. TOÇI**
ABSTRACT: Financing constraints have been one of the major impediments to doing busi-
ness in transition economies in general and South-Eastern Europe in particular. Utilizing 
firm-level survey data and extensive econometric modelling, the paper provides new evi-
dence on financing constraints, credit rationing and financing obstacles for firms in Alba-
nia, Bosnia and Herzegovina, Bulgaria, Croatia, Macedonia, Romania and Serbia and 
Montenegro. The findings suggest that these phenomena are prevalent in the SEE region, 
especially in the small business sector, a driving force of economic development in these 
countries. Based on the findings, a number of policy implications aiming at reducing finan-
cing constraints for the small business sector are derived.
KEYWORDS: credit rationing, f inancing constraints, South-East Europe
 
JEL classification: D92, G21
1. INTRODUCTION
The transformation of the banking sector in transition economies has been one of the 
most dynamic and challenging aspects of the transition process. Reforms in South East 
European countries (SEE) were long delayed, especially in the first decade of transition. 
The banking sector was relatively inefficient, with a weak supervisory capacity, and old 
lending practices paved the way on many occasions for severe crisis and low levels of 
financial intermediation (Anderson and Kegels, 1998). In the second decade of transi-
tion, however, the banking sector in SEE underwent a fundamental restructuring and the 
* Corresponding Author; Staffordshire University; Leek Rd.; Stoke-on Trent, ST4 2DF, United Kingdom;
e-mail: I.Hashi@staffs.ac.uk.
**Central Bank of the Republic of Kosovo and University of Prishtina; Garibaldi 33, 10000 Prishtina, Republic 
of Kosovo; e-mail: valentintoci@bqk-kos.org. The views expressed in this paper are those of the authors and 
do not reflect the views of the institutions that they represent.
The authors would like to thank Professor Jean Mangan and two anonymous referees for useful comments.
ECONOMIC AND BUSINESS REVIEW  |  VOL. 12  |  No. 1  |  201030
consolidation. Reforms of the macroeconomic environment and institutions of a market 
economy were undertaken, which promoted stability and the prospect of financial deep-
ening. Notwithstanding the progress, the level of banking sector credit to enterprises in 
transition economies in general and SEE in particular has not reached that of countries 
with comparable levels of economic development (EBRD, 2006). This is because banks 
have imposed credit rationing on creditworthy enterprises for a variety of reasons: inef-
ficiencies in banks themselves, information asymmetry between borrowers and banks, 
poorly functioning institutions such as the rule of law, and other institutional and mar-
ket failures. These concepts are well grounded in the theoretical literature of the last 
three decades (Stiglitz and Weiss, 1981; Fazzari et al., 1988; and others). Credit rationing 
of credit worthy firms has important adverse implications for growth of firms and the 
whole economy. Hence, it is important to investigate this phenomenon in the context of 
SEE countries where the evidence to date is scarce. 
In this paper, the availability of finance is examined by bringing together two strands of 
the literature (credit rationing and financing constraints) into a single empirical frame-
work, providing new evidence on credit rationing, financing constraints and financing 
obstacles faced by firms in SEE. The paper focuses on the effects of this phenomenon on 
specific groups of firms such as SMEs, state-owned and foreign-owned firms, firms with 
good prospects, etc. Given that small firms are the most dynamic sector of the economy 
and critical to economic growth in SEE countries, the impact of market imperfections 
limiting their access to external finance, and therefore their potential growth, is also 
addressed in the paper. Other firm characteristics such as ownership, past performance, 
financial disclosure, etc., that may influence the degree of financing constraints are also 
explored. Other working hypotheses, such as the persistence of soft budget constraints 
for state-owned enterprises, whether credit rationing is a supply- or demand-driven phe-
nomenon, the ability of banks to distinguish between ‘good’ and ‘bad’ borrowers, are 
also tested. Using a large survey of firms in Albania, Bosnia and Herzegovina, Bulgaria, 
Croatia, Macedonia, Romania and Serbia and Montenegro for 1999, 2002 and 2005, and 
extensive econometric modelling, the prevalence of financing constraints and credit ra-
tioning in the SEE region is confirmed, especially for the small business sector. However, 
during the period under review there has been a substantial decrease in financing con-
straints, especially in the category of small firms, implying a shift in the lending technol-
ogy of banks towards the small business sector. Finally, a number of policy suggestions 
aiming at reducing financing constraints for the small business sector are presented.
The paper is organized as follows. In the next section, theories of credit rationing and 
financing constraints and some of the empirical work in the area are critically examined. 
While much of this literature was developed in the context of mature market economies, 
there have been a number of attempts to apply these concepts to the more developed tran-
sition economies. Section 3 discusses the empirical work of this paper – the data, models 
and results. Section 4 concludes. 
I. HASHI, V. Z. TOÇI  |  FINANCING CONSTRAINTS, CREDIT RATIONING AND FINANCING... 31
2. REVIEW OF THE LITERATURE
2.1 Credit rationing and financing constraints: the theory 
There is a well established literature arguing that imperfections in the credit market stem 
from information asymmetries, transaction costs and agency issues. Such imperfections 
may give rise to the credit rationing – a state in which information asymmetries between 
lenders and borrowers may result in the equilibrium interest rate not clearing the mar-
ket. Instead, the demand for loans will exceed the supply and banks will deny credit to 
some borrowers who are observationally indistinguishable from those who receive loans 
despite their willingness to pay the prevailing interest rate. For example, if the borrower 
bears no cost when project returns are lower than the debt obligation, the moral hazard 
argument can lead to credit rationing because the borrower may divert the funds to more 
risky project ex post (Stiglitz and Weiss, 1981), engage in asset substitution (Schwartz, 
1981), exert an inappropriate degree of effort on the project (Aghion and Bolton, 1997; 
Ghosh, et al., 2001) or even falsely declare bankruptcy (Williamson, 1986). Banks can-
not distinguish these borrowers from better quality ones because of information asym-
metries, hence, they will ration their supply of credit. 
The information asymmetry theories of credit rationing have often been criticized for 
assuming that banks are unable to distinguish between borrowers – given that banks are 
in the information processing business and have specialized expertise in analyzing credit 
risk (Riley, 1987; Inderst and Muller, 2007).1 For example, Riley (1987) criticizes the work 
of Stiglitz and Weiss by arguing that as long as high risk and high return are positively 
correlated, the adverse effect of risk may be offset by a favourable effect of returns and, as 
the number of observationally distinct groups increases, credit rationing may not be an 
empirically important phenomenon. Milde and Riley (1988) develop the ‘bank screen-
ing hypothesis’ in which separating equilibria with no rationing is attained where banks 
screen their borrowers by offering larger loans to safer borrowers and by sorting out dif-
ferent risk classes. However, banks can separate small firms from large ones, firms in one 
sector from those in other sectors, etc. Since perfect screening is impossible, within each 
group some will receive loans while apparently identical firms will not and credit ration-
ing will still occur.
The magnitude of credit rationing depends on the extent to which information asym-
metries are more problematic for a specific group such as small firms, as opposed to 
other groups. If small firms have higher informational problems and are subject to credit 
rationing, it does not necessarily mean that they have ‘bad’ projects in hand. As Storey 
(1994) notes, if business proposals are turned down for reasons not connected with the 
1 Some authors refer to the advances in credit scoring as the evidence that banks can distinguish between 
customers of varying risk. However, credit scoring models used by commercial banks are statistical models 
based on borrowers’ observable characteristics. Notwithstanding various estimation biases in the credit scor-
ing models (see for example Greene, 1998; Parnitzke, 2005), the rating is made based on some average pattern 
of observationally distinct risk classes. Despite this, within each risk class, due to information asymmetries 
and the inability to sort borrowers perfectly, credit rationing may still emerge.
ECONOMIC AND BUSINESS REVIEW  |  VOL. 12  |  No. 1  |  201032
viability of the project itself, e.g., because firms lack track record and collateral or are 
small, then credit rationing may be a problem in the credit market and may become the 
subject of government intervention. As Cressy (1996) notes, although the theoretical un-
derpinnings of credit rationing theory may be challenged, its implications seem to have 
been accepted since in many countries substantial sums of public money have been spent 
on alleviating this problem. 
It is important to recognize that if firms have access to external finance, they may still 
limit their investments to internally generated funds and underinvest because informa-
tion and incentive problems may introduce a wedge between the costs of internal and ex-
ternal finance – as the financing constraints literature argues. Financing constraints refer 
to the inability of firms to finance their desired level of investment that they would have 
undertaken in a perfect capital market. In perfect capital markets firms are indifferent 
between what sources they use (internal or external) to finance their investment. They 
would not find it difficult to raise external finance when profitable investment opportu-
nities arise and, in this framework, internal and external funds are perfect substitutes. 
However, given market imperfections, reliance on internal funds may be higher since 
raising external finance is either more costly or impossible. The ‘pecking order theory’ 
(Myers and Majluf, 1984) or the ‘hierarchy of finance hypothesis’ (Fazzari et al., 1988) 
suggest that external finance will be more costly than internal funds for financing invest-
ment and that firms will not turn to external funds until internal sources are exhausted. 
This is due to a number of reasons, including monitoring costs (Townsend, 1979; Wil-
liamson, 1987), asymmetrical information entailing moral hazard and adverse selection 
(Jensen and Meckling, 1976; Stiglitz and Weiss, 1981), transaction costs in issuing bank 
debt, such as costs of application, screening costs, bankruptcy costs, etc. While the em-
pirical literature, almost without exception, refutes the case of perfect capital market, the 
work on credit rationing has produced ambiguous results. In order to assess the validity 
of ‘bank screening hypothesis’ and credit rationing if it is important problem, the dispute 
has to be resolved by testing the two hypotheses empirically. The present study finds that, 
in the context of transition economies, both theories may hold.
2.2 The empirical literature
The bulk of the empirical studies on credit rationing and financing constraints is based 
on the experience of firms in developed market economies. In more recent years there 
have been a number of attempts to apply the theory to the more advanced transition 
economies. However there has been almost no attempt to assess the nature or extent of 
credit rationing and financing constraints in the less advanced SEE countries. 
The empirical literature on credit rationing and financing constraints has developed in 
two distinct frameworks. In terms of credit rationing, as Jaffee and Stiglitz (1990) note, 
the magnitude of this problem in the economy could be measured if the demand and 
supply for credit was known. However, what is observed is the quantity of credit that 
is transacted and not the excess demand for credit. Hence, credit rationing may be ex-
tremely difficult to identify, thus empirical tests are mainly indirect (Parker, 2002). One 
test for credit rationing is when the data allows the researcher to distinguish between 
firms that have applied for credit and those that have not (the demand effect) and also 
between firms that have been successful or unsuccessful in their application for credit 
(the supply effect). Using enterprise survey data from the Czech Republic, Hungary and 
Poland in the mid-1990s which allowed this distinction, Bratkowski, et al. (2000) tried to 
determine how banks differentiate between different groups of firms and how do they 
allocate credit: do they lend to firms with higher profitability; do they provide credit to 
firms with good prospects but weak track records; or do they treat collateral availability 
as the most important determinant of offering a loan? The authors found that banks 
were able to identify successful firms by lending to more profitable ones even without a 
track record but which were able to provide collateral. They also found that firms with 
less profit were less likely to apply for loans (e.g., no evidence of adverse selection). Based 
on these results, the authors conclude that credit rationing was not pervasive in the three 
countries in the early years of transition.
Studying the credit rationing problem in the housing loans market in the U.S., Jappelli 
(1990) explores the degree of rejection of applicants for bank loans as a measure of credit 
rationing. However, the author identified another group, those who did not apply for 
loans because they perceived that their application will be rejected by the bank. This 
group may be considered as credit rationed too since they cannot be treated as having 
had no demand for loans. The exclusion of this group may lead to biased results, because 
the self-selection of applicants may induce banks to adapt screening rules that differ from 
those that would prevail if this group of borrowers were also to apply.2 Jappelli (1990) 
classifies this group as ‘discouraged’ borrowers and, together with the rejected group, 
considers them as ‘credit constrained’ borrowers to distinguish them from the strict defi-
nition of credit rationing. Similarly, Levenson and Willard (2000) adopt the definition of 
Jappelli (1990) for credit rationing and use survey data for investigating the small busi-
ness sector in the U.S. The authors conclude that credit rationing was not a pervasive 
phenomenon in the U.S. economy.3 
The financing constraints literature, on the other hand, holds that if firms can easily 
obtain external funds without paying a premium, their investment decisions will be less 
sensitive to internally generated funds. In contrast, when the premium is high, firms use 
internally generated funds to fund their investment first and external funds are sought 
only after their internal resources are exhausted. The empirical question in this line of 
inquiry is to investigate whether a firm’s investment decision is sensitive to changes in 
net worth, holding investment opportunities constant. In these models, investment is 
regressed on some proxies of net worth (e.g., cash flow or liquidity) and other firm char-
2 Besanko and Thakor (1987) point out that in some cases the bank’s credit policy discourages borrowers from
applying for credit by using non-price mechanisms, e.g., collateral or application procedures, and that the 
bank need not explicitly reject borrowers to induce them to exit the market.
3 However, contrary to this, the bulk of other empirical studies have found that credit rationing is indeed bind-
ing – Perez (1998) for the U.S. economy, Angelini and Generale (2005) for developing economies, and Rizov 
(2004) for a transition economy.
I. HASHI, V. Z. TOÇI  |  FINANCING CONSTRAINTS, CREDIT RATIONING AND FINANCING... 33
ECONOMIC AND BUSINESS REVIEW  |  VOL. 12  |  No. 1  |  201034
acteristics. If the coefficient of cash flow is positive and significant, then it is said that 
firms face financing constraints. In this framework firms are classified a priori as more 
or less likely to face financing constraints based on the researchers’ beliefs about the sys-
tematic differences between firms regarding their information opacity, riskiness, etc. For 
instance, if small firms are perceived to be more financially constrained than large ones, 
a common feature of almost all investigations is that the sample is divided into sub-sam-
ples of small and large firms and then the investment model is estimated for the two sub-
samples. If investments by small firms are more sensitive to internally generated funds, 
as measured by the absolute size of the relevant coefficient, it is suggested that these firms 
are more financially constrained than their larger counterparts.
There are many a priori criteria on the basis of which firms may be considered to be 
more or less financially constrained. For example, Gertler (1988), Gilchrist and Zakra-
jsek (1995) use the size and age of firms as the criteria for financing constraints. As they 
point out, ‘informational frictions’ that add to the financing costs, and hence financing 
constraints, apply mainly to younger firms with a short track record and with a high 
degree of idiosyncratic risk and firms that are likely to be not well collateralized. Small 
firms in general are more likely to share these attributes and may be perceived as more 
risky. Also the screening and monitoring of small firms may be proportionately costlier 
to the lender. From a bank’s point of view, fixed lending costs related to loan appraisals 
and monitoring make costs per dollar lent relatively higher for small firms (Saito and Vil-
lanueva, 1981). Hence, they would be more financially constrained than larger and older 
firms. Also from the firm’s perspective, transaction costs of the application process are 
relatively higher for small firms. All these attributes may lead small and young firms to 
face higher financing constraints. 
The ownership of firms (particularly state ownership and foreign ownership) have some-
times been used as the a priori criterion to distinguish between more and less financially 
constrained firms. In the transition context, one may expect that state-owned enterprises 
(SOEs) face less financing constraints because of the prevalence of soft budget constraints 
(SBCs) and the variety of government subsidies. Perotti and Carare (1996) and Konings, 
et al. (2002) have shown that in Bulgaria and Romania, SOEs did not exhibit financing 
constraints due to the prevailing SBC regimes despite being highly indebted and hav-
ing negative cash flows. While in developed market economies this finding would be a 
sign of perfect capital market, in the transition context it is seen as an indication of SBC. 
However, in later years of transition subsidies to SOEs were cut, often following IMF rec-
ommendations, and budget constraints on SOEs hardened. In addition, after the entry of 
foreign banks access to finance for SOEs was reduced especially because of their delayed 
restructuring. On the other hand, foreign-owned firms, especially those originating from 
developed countries, may be expected to face lower financing constraints. This is because 
of their ability to access financial markets in their home countries, obtain financial re-
sources through foreign direct investments, etc. The possible lower financing constraints 
faced by foreign firms may be explained by the fact that these firms also bring know-how 
and new technology, have better governance and may be more transparent to their lend-
ers, and thus, may face less financing constraints from financial markets in their host 
4 In the literature, the terms credit rationing, credit constraints, financing constraints, liquidity constraints 
and recently, financing obstacles are sometimes used interchangeably. In this paper ‘financing constraints’ 
will refer to whether firms can use their internal funds to finance their investment or have to rely on external 
funds (and to what extent) (section 3.3.1); ‘credit rationing’ will refer to the case of firms which, despite their 
ability to pay the cost of credit, cannot obtain credit and inolves more direct tests of credit rationing (section 
3.3.2) and ‘financing obstacles’ will refer to the firms’ self-reported financing obstacles for their operation and 
growth (section 3.3.3).
countries. Perotti and Vesnaver (2004), for example, show that foreign-owned firms in 
Hungary face less financing constraints. 
Another approach to the study of financing constraints has been recently developed 
by Beck, et al. (2006). Unlike previous studies that inferred financing constraints from 
company financial statements, these authors maintain that firms’ financing obstacles, 
which are indicative of financial constraints, can be identified directly by asking firms 
about these constraints. Using the survey data for 10,000 firms in 80 countries around 
the world in 1999, they test the severity of self-reported financial obstacles facing firms 
on the basis of firm and country characteristics. Given that the previous empirical work 
often relied on balance sheet and income statement data of listed companies, the Beck et 
al. approach opened up the possibility of extending these investigations to the smaller 
transition economies of SEE. In these countries stock markets are either in their infancy 
or non-existent; they are also dominated by small firms which are not listed and more in-
formationally opaque, or their financial statements are difficult or impossible to obtain. 
Hence, the survey data may be more appropriate for the analysis of financing constraints 
facing the SME sector. The large-scale Business Environment and Enterprise Perform-
ance Surveys (BEEPS), conducted by the World Bank and EBRD, offer the possibility of 
using this approach for the SEE countries where the evidence on financing constraints, 
credit rationing and financing obstacles of firms is scarce. 
3. FINANCING CONSTRAINTS, CREDIT RATIONING AND FINANCING OB-
STACLES IN SEE4
3.1 The empirical framework
The empirical work in the remainder of the paper utilizes a survey-based approach to 
identify the determinants of firms’ financing constraints. We use a database derived from 
three rounds of BEEPS in 1999, 2002 and 2005, focusing on the sub-set of firms in the 
SEE region. Following the approach employed by Beck, et al. (2006), and extending it in 
a number of ways, we investigate whether age, size and ownership have any influence on 
the various dimensions of firms’ perceived financing constraints. The study differs from 
Beck, et al. (2006) in many respects. First, the investigation is specific to transition coun-
tries in SEE. Second, the study extends the period of analysis using data from the 2002 
and 2005 BEEPS. Third, using both logit and ordered logit models, several dimensions 
of firm perceptions of financing constraints are considered: a general financing obstacle, 
specific obstacles such as high interest rates or high collateral requirements, access to 
long-term and short-term loans, etc. Fourth, more objective variables such as the share 
I. HASHI, V. Z. TOÇI  |  FINANCING CONSTRAINTS, CREDIT RATIONING AND FINANCING... 35
ECONOMIC AND BUSINESS REVIEW  |  VOL. 12  |  No. 1  |  201036
of investment that firms finance through internal funds and through bank loans are em-
ployed, bringing the estimation model closer to the Fazzari, et al. (1988). Fifth, additional 
control variables such as firm performance measures (profitability and sales growth) and 
whether or not firms use accounting standards are introduced. Finally, the analysis con-
tains more direct evidence of credit rationing by explicitly modelling the firm’s decision 
to participate in the credit market and the lender’s decision to reject or accept the firm’s 
application for a loan. This is possible because the 2002 and 2005 BEEPS provide more 
detailed information on financing issues.  
To assess the determinants of the likelihood of firms applying for a loan and being de-
nied credit (i.e., direct credit rationing), we use the approach adopted by Bratkowski, et 
al. (2000). However, unlike Bratkowski, et al. (2000) who used a logit model to estimate 
two separate equations (the likelihood for applying for a loan and the likelihood being 
denied the loan), we use the Heckman model which also enables us to deal with the pos-
sible selection bias. The model of ‘discouraged’ borrowers is also tested by employing a 
logit estimation procedure, following Jappelli’s (1990) and Levenson and Willard’s (2000) 
methodology. However, the model here differs from Jappelli’s approach in the sense that 
here, borrowers are firms and not households; whereas the difference of this method to 
the Levenson and Willard (2000) approach is that they calculate the estimates for dis-
couraged borrowers indirectly based on the parameter estimates of the regression of 
firms that apply for loans and are denied credit.5 In addition, using logit and ordered 
logit models, the firms’ self-reported difficulties in accessing short-term and long-term 
loans are assessed.6 One of the reasons for employing several dimensions of financing 
constraints is that the firm’s self-reported financing constraints are subjective by nature 
and may not represent the real importance of specific constraints. It is therefore useful to 
combine these with more objective indicators as explained above. This would also meet 
the need for ‘methodological cross-checking’ advocated by Charnes, et al. (1988).   
3.2 Data and Models 
The BEEPS is a joint World Bank/EBRD project designed to collect firm-level informa-
tion on the impact of business environment through detailed enterprise surveys at regu-
lar intervals. The surveys were conducted in transition economies in 1999, 2002 and 2005 
and covered about 4,000-9,000 firms. The survey includes a broad range of questions 
aimed at evaluating the nature of obstacles faced by firms in important areas such as in-
frastructure, judiciary, business regulation, crime, corruption, taxation and finance. The 
entrepreneurs were asked to assess and rank on a 1 to 4 scale how problematic a particular 
area (e.g. finance) is for the operation and growth of their firms. 
5 This is because their survey did not provide direct information on whether non-applicants expect to be 
denied, while the BEEPS provides direct information on the reasons why some firms did not apply for a loan, 
including firms’ expectation of denial.
6 This method of measuring credit rationing is similar to Hersch, et al. (1997) though the focus of their study 
was different – the previous business experience of firms’ owners and whether they were members of the 
nomenkatura.
The general model employed in the empirical framework utilizing different specifica-
tions and estimation methods can be represented as follows:
Yi  = α + βX1i + γX2i + εi
where Yi represents various measures of financing constraints; X1i the vector of variables 
representing different firm characteristics; X2i country dummies; α, β, γ parameters to be 
estimated; and εi the random error. In the following section we use several versions of this 
model and a variety of econometric estimation methods. 
A number of indicators have been used as measures of financing constraints (Yi). In sec-
tion 3.3.1, we use the proportion of a firm’s investment expenditure financed by the firm 
(i) internal funds and (ii) bank loans.7 Given the nature of the data (with both ratios vary-
ing between zero and 100 percent and with many corner solution observations at both 
ends, the Tobit model (‘censored regression model’) is used for the purpose of estimation 
as the most appropriate model.8 This approach, as with Fazzari, et al. (1988), posits that 
firms which rely more on internal funds, as opposed to external funds, to finance their 
investment are more financially constrained. In section 3.3.2, we focus on the probability 
of a firm applying for a loan and the probability of it being denied the loan and assess the 
factors influencing these probabilities. We also assess the model of discouraged borrow-
ers and the determinants of the difficulties of firms’ access to short-term and long-term 
loans. In section 3.3.3, we assess the determinants of not only broad financing obstacles 
but also specific obstacles such as high interest rates and collateral requirements. 
Firm characteristics in all models include the firm’s age (logged), size (measured by dum-
my variables indicating small and medium compared to the base category of large firms), 
ownership, performance (for example sales growth or profitability9 as proxies for credit 
worthiness or future prospects), accounting method used (to signal transparency of fi-
nancial reporting), sector and country of origin. It is expected that younger and smaller 
firms to be more financially constrained, e.g., to finance a higher proportion of their 
investment through internal funds, and a smaller proportion through bank loans, com-
pared to their older and larger counterparts. Majority foreign ownership are expected 
to be less financially constrained than private domestic companies. Firms with better 
7 There are also other sources of funds used by firms to finance their investments such as loans from family 
and friends. Given that these two sources of funds best demonstrate the presence of constraints in the context 
of the investigation, other sources of funds used by firms to finance their investments are excluded from the 
analysis.
8 The tobit model is applicable to censored outcomes, i.e., situations in which the dependent variable is not ob-
servable for part of the population and corner solution outcomes, i.e., situations in which the dependent vari-
able is an observable outcome but the optimal choice will be a corner solution. For corner solution outcomes, 
it makes more sense to call the resulting model a ‘corner solution model’. Unfortunately, the name ‘censored 
regression model’ appears to be firmly entrenched in the literature (see Wooldridge, 2001).
9 Because of the nature of the dataset and the fact that some questions have changed from one round of survey 
to another, two profitability variables are used in the estimations: Profitability1 for the 2002 survey, a dummy 
variable taking the value of 1 if the firm had positive gross profit to total sales ratio in 2001 and 0 otherwise; 
and Profitability2 for the 2005 survey, a dummy variable taking the value of 1 if the firm had reinvested some 
of its 2003 profits in 2004, and 0 otherwise.
I. HASHI, V. Z. TOÇI  |  FINANCING CONSTRAINTS, CREDIT RATIONING AND FINANCING... 37
ECONOMIC AND BUSINESS REVIEW  |  VOL. 12  |  No. 1  |  201038
performance and prospect may be better placed to use their own funds and also apply 
for bank credit. Firms with better prospects may be in need of more external finance 
in order to exploit investment opportunities; hence, may be more likely to apply for ex-
ternal finance. To the extent that banks are able to identify this, they may be more will-
ing to provide credit, thus, firms may be less constrained. At the same time, firms with 
low profitability may face liquidity problems and turn to banks for funds. This may be 
a sign of adverse selection allowing poorly performing firms to enter the loan market, 
worsening the pool of applicants which, in turn, may induce banks to limit credit avail-
ability. Profitability may decrease the likelihood of a firm participating in the credit mar-
ket given the stringent requirements by banks and high costs of external finance. Thus, 
profitable firms do not turn to banks for funding (the hierarchy of finance hypothesis). 
On the other hand, to the extent that firm profitability as an indicator of creditworthi-
ness is observable to the bank, viable firms would find it easier to access external finance.
The three BEEP surveys differ from each other in a number of ways, particularly in terms 
of some questions. This means that the model run for each year, therefore, is slightly 
different from those run for other years because it may contain one or two different in-
dependent variables. But in addition to regressions for each year of the survey, the results 
also include a pooled regression which includes variables which have been covered in 
all three surveys. The pooled regressions contain year dummies which explain whether 
the financing constraints have been relaxed over time given the dynamic changes in the 
banking sector in the SEE countries. 
Table 1 summarises the descriptive statistics of the variables used in the models discussed 
in the next section. The precise definition of variables is presented in Appendix 1. As Ta-
ble 1 shows, most variables (except for Internal Funds, Bank Loans, Age and Sales Change 
which are continuous) are either ordered or dummies. From the summary statistics, it 
can be observed that, on average, firms rate high interest rates as a greater obstacle in the 
three rounds of BEEPS compared to other dimensions of financing obstacles. Different 
dimensions of financing obstacles are significantly correlated, with the coefficients rang-
ing from 0.29 to 0.66 in various years. Interestingly, firms’ self-reported obstacles are not 
correlated with more objective measures such as Internal Funds and Bank Loans.10
10 The correlation matrix of variables is available upon request.
Table 1: Descriptive statistics of variables
Variable Year Obs Mean Std Min Max
Dependent Variables 
Internal Funds 1999 820 62.17 39.10 0 100
2002 1430 60.27 37.96 0 100
2005 2000 61.47 36.25 0 100
Bank Loans 1999 820 8.72 21.18 0 100
2002 1430 9.46 20.99 0 100
2005 2000 13.02 23.66 0 100
Difficulty accessing short-term loans 1999 na na na na na
2002 1293 2.72 2.13 1 5
2005 na na na na na
Difficulty accessing long-term loans 1999 na na na na na
2002 1260 3.07 2.14 1 5
2005 na na na na na
Discouraged borrowers 1999 na na na na na
2002 na na na na na
2005 1426 0.32 0.47 0 1
Aplicants 1999 na na na na na
2002 na na na na na
2005 2039 0.52 0.50 0 1
Rejected borrowers 1999 na na na na na
2002 na na na na na
2005 1021 0.05 0.22 0 1
Financing obstacle 1999 855 3.09 1.10 1 4
2002 1362 2.42 1.19 1 4
2005 1949 2.35 1.14 1 4
High interest rate obstacle 1999 817 3.41 0.99 1 4
2002 1410 2.67 1.12 1 4
2005 1968 2.62 1.13 1 4
Collateral obstacle 1999 734 2.58 1.24 1 4
2002 na na na na na
2005 na na na na na
Access to long-term loan obstacle 1999 752 3.22 1.13 1 4
2002 na na na na na
2005 na na na na na
Explanatory Variables
Age 1999 845 16.14 22.56 1 172
2002 1464 16.28 19.73 3 202
2005 2040 17.84 20.65 4 180
Small 1999 873 0.58 0.49 0 1
2002 1464 0.66 0.48 0 1
2005 2040 0.68 0.47 0 1
Medium 1999 873 0.22 0.41 0 1
2002 1464 0.20 0.40 0 1
2005 2040 0.22 0.41 0 1
Large 1999 873 0.21 0.41 0 1
2002 1464 0.15 0.36 0 1
2005 2040 0.11 0.31 0 1
State 1999 873 0.22 0.41 0 1
2002 1349 0.14 0.36 0 1
2005 2016 0.10 0.29 0 1
Private domestic 1999 873 0.71 0.45 0 1
2002 1349 0.75 0.46 0 1
2005 2016 0.80 0.40 0 1
Foreign 1999 873 0.08 0.26 0 1
2002 1349 0.11 0.36 0 1
2005 2016 0.11 0.31 0 1
Sales change 1999 812 12.61 59.32 -90 700
2002 1360 18.31 58.22 -90 600
2005 1968 10.63 36.92 -90 310
Profitability1 2002 1375 0.90 0.30 0 1
Profitability2 2005 1939 0.84 0.37 0 1
Accounting standards 1999 788 0.56 0.50 0 1
2002 1294 0.49 0.50 0 1
2005 1972 0.60 0.54 0 1
Note: a) Sector and country dummies not reported.
I. HASHI, V. Z. TOÇI  |  FINANCING CONSTRAINTS, CREDIT RATIONING AND FINANCING... 39
ECONOMIC AND BUSINESS REVIEW  |  VOL. 12  |  No. 1  |  201040
3.3 Results
3.3.1 Financing Constraints: Reliance on Internal Funds and Bank Loans
The Tobit model is used to assess the determinants of the proportion of firms’ invest-
ments financed through internal funds and bank loans (two separate specifications).11 
The Tobit estimates are presented in Table 2 (specifications 1–8). For each dependent 
variable, there are four specifications (one for each of the three years of survey and one 
for the pooled data). The estimated coefficient on age is positive but insignificant in all 
but one specification. The exception is specification 4 with internal funds as the depend-
ent variable, indicating that older firms finance a larger proportion of their investment 
through internal funds, i.e., are more constrained, which is contrary to expectations and 
what is generally perceived in the literature.12 However, age is insignificant in the re-
gressions with bank loans as the dependent variable. The change in the structure of the 
banking sector in the region (e.g., the entry of foreign banks and the closure of domestic 
banks) may also be an explanation for this outcome, in the sense that all firms were ‘new’ 
to foreign banks and the relationships had to be established from scratch.13 
The coefficients on size (especially small firms) are in most specifications highly signifi-
cant and have the expected sign, indicating that SMEs may rely more on internal funds 
and less on bank loans to finance their investment needs, compared to larger firms – in 
few specifications the coefficients are insignificant at conventional levels. The estimated 
coefficient on state ownership is negative and significant in all the specifications for bank 
loans, indicating that the SOEs finance a smaller proportion of their investment through 
banks. This provides some evidence that hardening of budget constraints, or even credit 
rationing, has become binding for SOEs. However, they do not seem to rely more on 
internal funds. This indicates that SOEs may still be receiving subsidies from the govern-
ment, but not the ‘soft’ lending from the banks as in the early transition years.
The effect of foreign ownership is rather unclear. The variable is significant at five per-
cent level in only one specification, indicating that foreign firms rely less on bank loans 
than domestic firms to finance their investments. The evidence from the ‘internal funds’ 
specification suggesting that these firms do not rely more on internal funds to finance 
their investment, may support the view that foreign firms may turn to banks in their 
11By comparing the results with a similar probit model, where the dependent variable is a dummy taking the 
value of one if any percentage of investment is financed by internal funds (or bank loans) and zero otherwise, 
it can be shown that there is no misspecification in terms of heteroscedasticity and non-normality. This is 
because the probit parameter estimates are broadly similar in magnitude to Tobit parameter estimates divided 
by the estimated standard error of the regression – though they cannot be the same due to sampling error (see 
Wooldridge, 2001). The results are available upon request.
12 A specification replacing the age variable with a dummy taking the value of one if the firm is less than five 
years old and zero otherwise (not reported here), produces statistically insignificant results for all specifica-
tions.
13 Another explanation may be that the definition of this variable is not consistent across years: in the 1999 
survey firms were asked about the year when they started their business, while in 2002 and 2005, when they 
started business in particular country. Furthermore, the later two rounds of BEEPS exclude firms younger 
than four years from the sample, i.e., new entrants, which presumably face the highest constraints.
home countries or their parent banks for funds (presumably with better conditions). Re-
garding the performance variables, sales growth is estimated to have a positive effect, 
but is significant in only two specifications, providing some indication that firms with 
good prospects use both internal funds and bank loans more than firms with weak pros-
pects. Profitability has a positive sign and is statistically significant at the one percent 
level suggesting that more profitable firms finance a greater proportion of their invest-
ment through internal funds. However, this variable is not significant in the specification 
with Bank Funds. This may support the ‘hierarchy of finance hypothesis’ suggesting that 
because of the premium with external funds, firms finance their investments internally 
and do not turn to banks for finance until internally generated funds are exhausted.
The use of accounting standards has a statistically significant impact on the proportion 
of investment that firms finance through bank loans in two specifications. This suggests 
that firms with clearer financial reporting were able to rely more on banks to finance 
their investment. This variable has no effect on the use of internal funds (or only margin-
ally as in one specification). 
I. HASHI, V. Z. TOÇI  |  FINANCING CONSTRAINTS, CREDIT RATIONING AND FINANCING... 41
ECONOMIC AND BUSINESS REVIEW  |  VOL. 12  |  No. 1  |  201042
Table 2: Financing constraints of firms in SEE
(1) (2) (3) (4) (5) (6) (7) (8)
Age 11.77 4.76 1.45 7.43* 0.06 10.10 2.82 4.64
(7.95) (6.71) (5.31) (4.10) (8.81) (7.77) (5.40) (4.31)
Small 37.78*** 21.13*** 7.25 19.98*** -52.04*** -40.17*** -18.18*** -32.33***
(10.46) (6.94) (5.58) (4.52) (11.66) (7.91) (5.61) (4.61)
Medium 21.67** 13.91** 4.55 9.82** -9.14 -17.17** -3.76 -8.76**
(9.67) (7.08) (5.56) (4.54) (9.89) (7.69) (5.48) (4.48)
State 6.37 0.52 -7.45 -1.63 -19.62* -30.63*** -20.31*** -26.91***
(9.60) (6.73) (5.68) (4.53) (10.25) (8.21) (6.09) (4.91)
Foreign 3.21 -3.51 7.59 1.80 -16.15 2.13 -12.26** -6.76
(12.72) (5.97) (4.96) (4.31) (14.91) (6.93) (5.11) (4.49)
Sales change 0.04 0.06 0.06 0.07*** -0.05 0.04 0.09** 0.02
(0.05) (0.04) (0.04) (0.02) (0.07) (0.04) (0.039) (0.02)
Profitability1 31.31*** 9.25
(6.92) (8.44)
Profitability2 5.25
(4.42)
Account. Standards -13.64* -0.94 -3.72 -0.25 2.32 8.53 7.90*** 5.35*
(7.62) (4.36) (3.08) (2.92) (9.11) (5.17) (3.04) (3.04)
Year 1999 5.30 -13.77***
(4.08) (4.37)
Year 2002 -9.49*** -5.00
(3.72) (3.91)
Constant 6.41 -37.25** 58.46*** 13.01 2.64 -6.65 -23.43 -11.43
(32.62) (17.09) (13.32) (11.52) (35.47) (20.37) (14.50) (12.72)
Obs. 674 1030 1845 2769 674 1030 1756 2769
Left Censored obs. 117 170 226 434 515 762 1135 1938
Uncensored obs. 306 536 1009 1436 143 249 596 778
Right censored obs. 251 324 610 899 16 19 25 53
Log Likelihood -2069.87 -3381.39 -6311. 35 -9235.35 -1014.52 -1683.85 -3798.00 -5181.86
LR chi2 52.18*** 201.44*** 117.27*** 182.99*** 71.50*** 143.52*** 151.58*** 279.94***
Pseudo R2 0.02 0.03 0.02 0.02 0.04 0.05 0.02 0.03
Notes: ***, **, * denote significance at 1%, 5% and 10%, respectively; standard errors in parentheses. 
19
99
20
02
20
05
Po
ol
ed
 
Internal Funds [Tobit model] Bank Loans [Tobit model] 
19
99
20
02
20
05
Po
ol
ed
 
In terms of the year effect, the dummy for 2002 is negative and statistically significant at 
five percent level indicating that firms in 2002 relied less on internal funds compared to 
2005. On the other hand, in the specification for bank loans, the year dummy for 1999 
is negative and statistically significant at the one percent level suggesting that banks in-
creased their role in supporting firms to finance their investment over the period. Over-
all, the estimates suggest that smaller firms and SOEs face the highest financing con-
straints. Profitable firms finance a larger proportion of their investment through internal 
funds, but not bank financing which seem to be in line with the pecking order theory. It 
also seems that, over time, access to bank loans has improved for firms in SEE.14
3.3.2. Testing for credit rationing: rejected and discouraged applicants
Table 3 shows the results of the estimations aimed at assessing more directly whether 
firms in the SEE region experience credit rationing by considering three different indica-
tors of credit rationing. First, the determinants of the likelihood of applying for a loan 
and being refused a loan are assessed. Second, the model of discouraged borrowers is 
employed. Third, the determinants of the likelihood of the firm reporting difficulties in 
accessing short-term and long-term loans are assessed.
i. The likelihood of applying for a loan and being refused
The Heckman model is utilized to assess the determinants of the likelihood of applying 
for a loan and being refused the loan (specifications 1 and 2). This model with partial 
maximum likelihood estimation is employed specifically to deal with the selection bias 
problem given that those who apply may systematically differ from those who do not ap-
ply, and the latter group may be influenced by and reflect a response to credit denials. In 
other words, because applying may be systematically correlated with unobservables that 
affect the likelihood of being denied, using only those who apply may produce biased 
estimates in the equation of the likelihood of being denied. For example, firms, including 
non-applying firms, may anticipate the bank screening procedures and the acceptance 
criteria, a situation which will result in self-selection in applying for credit based on the 
perception of success and conditions of a loan offer. Hence, it is likely that the sample is 
subject to selection bias. 
14 There is no systematic difference among country and sector control variables in any of the models used in 
the analysis. Given that the focus of the study is on firm characteristics and in order to present more clearly the 
variables of interest, sector and country specific control variables are not reported, although they are included 
in all regressions.
I. HASHI, V. Z. TOÇI  |  FINANCING CONSTRAINTS, CREDIT RATIONING AND FINANCING... 43
ECONOMIC AND BUSINESS REVIEW  |  VOL. 12  |  No. 1  |  201044
The first-stage equation includes the ‘Financing Obstacle’ variable as the part of selection 
since it may have an impact on the firm’s decision to participate in the credit market but 
not have an impact on the bank decision to lend.15 The coefficients in the first step equa-
tion show which factors influence the firm’s decision to apply for a loan while those in the 
second step show which factors influence the bank’s decision to deny a loan, given that 
the firm has applied for a loan. 
In the first specification estimating the likelihood of applying for a loan the estimates 
show that small and medium-sized firms are significantly less likely to apply for bank 
loans compared to their larger counterparts. This may indicate self-selection in the mar-
ket, i.e., those with little collateral and/or facing high transaction costs of the application 
process are less likely to enter the loan market. In terms of ownership, both SOEs and 
foreign firms seem less likely to apply for bank loans compared to private domestic firms. 
An explanation may be that SOEs do not apply because of the prospects of being denied; 
while foreign-owned firms do not apply due to the possibilities of receiving funds from 
their home countries/parent companies. Sales growth is positive and statistically signifi-
cant at the one percent level suggesting that firms with good prospects, i.e., those in need 
for more investment, seem to be more likely to apply for bank loans. This may indicate 
that adverse selection may not be a problem since firms with poor prospects are less likely 
to participate in the credit market and worsen the pool of applicants. This may also be 
an additional evidence of self-selection in the credit market in SEE. However, the profit-
ability proxy is positive but insignificant. Firms employing accounting standards seem 
to have a higher likelihood of applying for a loan. The selection variable (Financing Ob-
stacle) is positive and highly significant, indicating that firms reporting higher financing 
obstacles to their operation and growth are more likely to turn to banks for loans.
15 To correct for the sample selection bias, there needs to be a variable that affects the selection equation with-
out affecting the second. In the Heckman estimation, in the selection equation, the inverse Mill’s ratio (λ) is 
estimated for each observation. In all the cases with this type of data, the two-stage estimation procedure is 
necessary to calculate the Mill’s λ in order to infer whether the sample is subject to selection bias (see Heck-
man, 1979; Wooldridge, 2001 for the methodology). Mill’s λ is statistically significant, which suggests that the 
sample is subject to selection bias. Therefore, the Heckman method is the appropriate technique for estimating 
the second-stage equation of the likelihood of the bank decision to deny a loan.
Table 3: Tests for credit rationing of firms in SEE in 2005
(1) (2) (3) (4) (5)
Age 0.150 -0.020 -0.162 0.258 -0.005
(0.120) (0.028) (0.260) (0.268) (0.245)
Small -0.708*** 0.050** 1.735*** 1.056*** 1.323***
(0.131) (0.022) (0.326) (0.295) (0.270)
Medium -0.213* 0.014 0.714** 0.043 0.313
(0.130) (0.018) (0.326) (0.315) (0.280)
State -0.570*** 0.016 0.670** 0.208 0.346
(0.132) (0.030) (0.289) (0.262) (0.242)
Foreign -0.264** -0.031* -0.180 -0.389 -0.485**
(0.109) (0.016) (0.268) (0.254) (0.227)
Sales change 0.002*** -0.0002* -0.004** -0.005*** -0.004**
(0.000) (0.0001) (0.002) (0.001) (0.002)
Profitability1 -0.686** -0.455*
(0.269) (0.248)
Profitability2 0.072 -0.054** -0.673***
(0.094) (0.026) (0.182)
Account. Standards 0.276*** -0.006 -0.452*** -0.008 0.074
(0.068) (0.011) (0.147) (0.170) (0.155)
Financing Obstacle 0.148***
(0.030)
Constant -0.250 0.147* -1.103 -2.216*** -1.706***
(0.323) (0.075) (0.668) (0.671) (0.617)
Obs 1265 963 947
Log Likelihood -682.07 -493.81 -565.37
LR chi2 220.62*** 147.47*** 115.81***
Wald chi2
Mill's λ
Pseudo R2 0.139 0.13 0.10
Notes: ***, **, * denote significance at 1%, 5% and 10%, respectively; standard errors in parentheses. 
1684
-930.84
244.00***
-0.031***
Th
e l
ike
lih
oo
d o
f a
pp
lyi
ng
 fo
r 
a l
oa
n: 
1=
ap
ply
, 0
=o
the
rw
ise
    
 
(S
ele
cti
on
 M
od
el)
Th
e l
ike
lih
oo
d o
f b
ein
g 
rej
ec
ted
 a 
loa
n: 
1=
rej
ec
ted
, 
0=
oth
erw
ise
 (S
ec
on
d S
tag
e 
M
od
el)
Th
e l
ike
lih
oo
d b
ein
g r
eje
cte
d 
an
d d
isc
ou
rag
ed
 fr
om
 
ap
ply
ing
 fo
r a
 lo
an
=1
; 
Ot
he
rw
ise
=0
 (L
og
it M
od
el)
Im
po
sib
le/
ve
ry
 di
ffi
cu
lt 
Ac
ce
ssi
ng
 Sh
or
t-T
erm
 lo
an
s 
[L
og
it M
od
el]
Im
po
sib
le/
ve
ry
 di
ffi
cu
lt 
Ac
ce
ssi
ng
 L
on
g-
Te
rm
 lo
an
s 
[L
og
it M
od
el]
Heckman MLE model
The model of 
discouraged 
borrowers
Access to short-term and long-term loans
In the equation estimating the likelihood of being denied credit the dummy variable for 
small firms is statistically significant at the five percent level. The results indicate that it is 
small firms that face the highest likelihood of being denied credit in the SEE. The dummy 
variable for foreign firms is negative and statistically significant (at ten percent level) 
indicating that foreign firms are less likely to be denied credit. The coefficients on sales 
change and profitability have an estimated negative effect and are statistically significant. 
Consistent with the ‘bank screening hypothesis‘, this may be interpreted as banks being 
I. HASHI, V. Z. TOÇI  |  FINANCING CONSTRAINTS, CREDIT RATIONING AND FINANCING... 45
ECONOMIC AND BUSINESS REVIEW  |  VOL. 12  |  No. 1  |  201046
able to distinguish ‘good‘ from ‘bad‘ firms and firms with good prospects are less likely 
to be denied credit. It seems that banks do screen their applicants’ creditworthiness and 
viable firms are more likely to receive loans. However, this should be interpreted with a 
caveat since, inter alia, keeping profit, sales change and the use of some accounting stand-
ards constant, small firms are still less likely to participate in the credit market and more 
likely to be denied a loan. This may indicate that some credit rationing is present in the 
market from both the demand side (which seems more prevalent) and supply side, and 
that small firms face the highest constraints. Calculations from the BEEPS data reveal 
that 93.5% of firms that applied obtained the loans they sought. As a result, credit ration-
ing seems to work more through self-selection of firms on the demand side. 
ii. A model of discouraged borrowers
As discussed previously, it is inappropriate to treat firms which do not apply for loans as 
having no demand for loans. Discouraged borrowers who do not apply for a loan because 
of various reasons should be taken into account too (of all firms in the SEE sample only 
51.6 percent of firms applied for bank loans). The BEEPS 2005 dataset provides direct evi-
dence of reasons why firms may not apply for loans. Almost 60 percent of non-applying 
firms did not apply because they had no need for loans, the rest either considered inter-
est rates too high (11.3 percent), collateral requirements too high (2.8 percent), the pro-
cedures for the application complicated (6.3 percent), informal payments required (0.2 
percent), did not expect the loan will be approved (1.9 percent), a combination of above 
reasons (17.1 percent) and others (2.2 percent). Apart from firms that did not need loans 
and considered interest rates to be high, the rest (30.6 percent of non-applying firms) may 
be considered as ‘discouraged‘ borrowers, which jointly with the group of rejected ap-
plicants constitute the sample of credit rationed firms as defined by Jappelli (1990). The 
data also show that from the sample of applicants, 6.5 percent were denied credit while 
the remaining 93.5 percent received the loans they sought.16 
In the SEE the rejection rate is higher for small firms – roughly, one in ten small firms 
are denied credit, while the ratio for medium and large firms is one in twenty. However, 
if the fact that some firms may not be creditworthy is taken into account, then the credit 
rationed firms must account for a smaller proportion. Two performance measures were 
calculated for firms that were denied credit – the proportion of firms that were profit-
able and the proportion with growing sales. Around 67 percent of rejected firms were 
profitable and 43 percent had positive sales growth. Therefore, some creditworthy firms 
appear to be denied credit indicating that some credit rationing is present in the market. 
In the total sample of firms in the SEE, the rejected and discouraged firms, constituting 
the upper bound of credit rationing, were around 17 percent which is almost three times 
16 In a sample of small firms only, the corresponding figure for the U.S. economy in 1988 was 2.14 percent (Lev-
enson and Willard, 2000) and for Italy it was 2.7 percent during the economic boom in 1988 and 12.8 percent 
during the recession in 1993 (Guiso, 1998). In a more conservative approach, using only a sample of start-up 
firms in the U.K., Cressy (1996) found that only six percent of business start-ups were denied a loan.The results 
may be very different in developing countries. Bigsten, et al. (2000) find that more than half of the firms in the 
sample of six African countries had no demand for credit. Of those with a demand, only a quarter obtained a 
loan and small firms were constrained more.
more than that calculated by Levenson and Willard (2000) for the U.S. economy.17 Unfor-
tunately, due to data unavailability, it was not possible to assess the share of these firms in 
total employment and sales of the sample. 
In the model of discouraged borrowers, credit rationed firms are identified by a dummy 
variable taking the value of one if the firm was refused a loan or if it was a ‚discouraged 
borrower’, and zero otherwise. Table 3 (specification 3) presents the logit estimates of 
the likelihood of a firm being credit rationed. As in most previous models, age is insig-
nificant. The results show that SMEs are significantly more likely to be credit rationed 
than their larger counterparts. In this specification the coefficient on foreign firms is 
negative but not significant, while that for SOEs is positive and statistically significant at 
one percent level, indicating that SOEs are more likely to be credit rationed. Firms’ good 
prospects, as proxied by the change in the firm’s sales, have a negative and statistically 
significant effect on the likelihood of credit rationing. Profitability is also statistically 
significant at one percent level, indicating that more profitable firms are less likely to be 
credit rationed. The ‘use of accounting standards’ enters with a negative sign and statisti-
cally significant at one percent level suggesting that if firms are able to provide proper 
information to their lenders, the likelihood of being credit rationed will be reduced. In 
all, the results broadly suggest that being large, performing well and using some account-
ing standards reduce the likelihood of being credit rationed.
iii. Difficulties of obtaining short-term and long-term loans
We consider the determinants of the likelihood of the firm reporting difficulties in ob-
taining short-term and long-term loans on a 1 to 5 scale, with 1 being very easy and 5 
impossible, using both the logit and ordered logit models. Some studies, e.g., Hersch, et 
al. (1997) and Batra, et al. (2002), transform the ordered dependent variable: ‚how dif-
ficult is it to get a long-term loan from a bank‘ ranking from 1, very easy, to 4, very dif-
ficult, into a dichotomous dummy variable (by amalgamating the top two rankings into 
1 and bottom two rankings into 0) and employ the logit model. Using BEEPS data from 
2002, in this section the group of firms facing the highest difficulties in accessing loans 
is modeled using logit estimation. A dummy dependent variable is utilized taking value 
of one if the firm reports that it is impossible or very difficult to access short-term and 
long-term loans, and zero otherwise (Table 3, specification 4 and 5). In order to pick up all 
the variation in the data, an ordered logit model is also estimated for a dependent variable 
ranging from 1, very easy, to 5, impossible, indicating the difficulty experienced by firms 
in accessing loans. The results for an ordered logit model are largely consistent with logit 
estimates and are presented in Appendix 2. 
17 It should be noted that the approach taken here may result in some underestimation as the work of Levenson 
and Willard (2000) focused on a sample of small businesses only. However, it may also represent an overes-
timation since unlike Jappelli (1990) and Levenson and Willard (2000) that take into account only firms that 
expect to be turned down, here all reasons for non-application are taken into account – apart from ‘no need 
for loans’ or ‘interest rates considered too high’. The group of firms that consider interest rates to be high are 
excluded in order to approximate more closely the definition of credit rationing, i.e., the demand for loans 
will exceed the supply at prevailing interest rates. However, the results are basically the same if this group is 
included in the regression analysis.
I. HASHI, V. Z. TOÇI  |  FINANCING CONSTRAINTS, CREDIT RATIONING AND FINANCING... 47
ECONOMIC AND BUSINESS REVIEW  |  VOL. 12  |  No. 1  |  201048
As in most of previous specifications, age is insignificant. The reporting of greater dif-
ficulties in accessing either short- or long-term loans is statistically significant at one 
percent level for the group of small firms. However, the coefficient on medium-sized 
firms is insignificant. State-owned firms do not report significantly higher difficulties 
in accessing either long- or short-term loans, although the estimated coefficients have a 
positive sign. Being a foreign-owned firm decreases the likelihood of reporting difficul-
ties in accessing long-term loans compared to private domestic firms and the estimated 
effect of sales change and profitability are negative and statistically significant in both 
specifications. This indicates that firms with good prospects are less likely to be credit 
rationed. The results also suggest that firms using some accounting standards have no 
statistically significant effect on difficulties in accessing short- and long-term loans for 
the sample of firms in 2002. However, as shown previously, in later years (with 2005 data) 
these firms financed a greater proportion of their investment through bank funds and 
were less likely to be credit rationed, suggesting the financing environment for firms with 
better information availability has improved.   
3.3.3 The importance of financing obstacles for firms’ operation and growth
In this section the dependent variables indicate the firm’s perceived severity of several 
financing obstacles for the operation and growth of their business on a 1 to 4 scale: gen-
eral financing obstacle (Financing Obstacle), high interest rates (High Interest Rates Ob-
stacle), high collateral requirements (Collateral Obstacle) and access to long-term loans 
(Long-term Loan Obstacle). To better discriminate firms that report various dimensions 
of finance as a major obstacle, a logit model is employed for the dependent variable taking 
one if firm reports finance as a major obstacle and zero otherwise. As previously, the or-
dered logit models are also employed so that all the variation in the data is picked up. The 
estimates from the ordered logit models are broadly consistent with the logit estimates 
and are presented in Appendix 2. 
i. General financing obstacle
In this section the determinants of the general financing obstacle are assessed, using the 
three rounds of BEEPS and a pooled regression. Results are presented in Table 4 (speci-
fications 1–4). The estimates indicate that age generally does not have explanatory pow-
er for firms reporting finance as a higher obstacle. The estimated coefficient on small 
firms is positive and highly significant in all specifications indicating that small firms 
face greater financing obstacles compared to larger firms. The estimated effect of being 
a medium-sized firm is positive and significant in 1999 and in the model with pooled 
data. This provides some evidence that the medium-sized firms too face greater financ-
ing obstacles than their larger counterparts. Regarding the firm ownership, the dummy 
variable for SOEs is not significant in any specification (except in specification 1 at ten 
percent level). This does not support the assertion that SOEs may have had access to 
‘soft‘ lending, hence may report lower financing obstacles. The dummy variable for for-
eign-owned firms is negative and statistically significant in all specifications (except in 
specification 2). This may support the previous claim that foreign firms may have lower 
financing constraints because they have better governance and banks are more willing to 
lend to this group, they may have access to finance from banks in their parent countries, 
or foreign direct investments from their parent companies.
Regarding other explanatory variables, sales change is estimated to have a negative and 
statistically significant effect in nearly all specifications. This supports the previous find-
ing that firms with good prospects may face lower financing obstacles. Profitability has 
an estimated negative sign and is statistically significant at the one percent level in 2005, 
but in 2002 is insignificant. It is important to note here that profitable firms may re-
port lower financing obstacles either because this implies creditworthiness which may 
be identified by the lenders and have lower difficulties in accessing external finance, or 
they may report lower obstacles because their profits suffice for the needed investments 
to be undertaken. As shown previously, profitable firms do not finance a higher propor-
tion of investment through bank loans and are not more likely to apply for bank loans. 
However, the evidence suggests that if they turn to banks for finance, they are less likely 
to be denied. The variable indicating if the firm uses some accounting standards is not 
significant in any of the specifications. In terms of the year effect, in the regression with 
pooled data, the coefficients for 1999 and 2002 are statistically significant and positive. 
This suggests that the financing obstacles seem to have decreased over time for the firms 
in SEE (more so in the 1999-2002 period than in the 2002-2005 period), which is in line 
with the deepening of the financial markets in these countries.
 
I. HASHI, V. Z. TOÇI  |  FINANCING CONSTRAINTS, CREDIT RATIONING AND FINANCING... 49
ECONOMIC AND BUSINESS REVIEW  |  VOL. 12  |  No. 1  |  201050
 Table 4. Financing obstacles of firms in SEE
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Age -0.161 0.264 0.230 0.199 0.147 0.082 0.403* 0.411*** -0.066 -0.051
(0.200) (0.251) (0.225) (0.137) (0.238) (0.235) (0.207) (0.134) (0.229) (0.217)
Small 0.651** 0.584** 0.650*** 0.620*** 0.635** 0.360 0.637*** 0.485*** 1.191*** 1.115***
(0.267) (0.272) (0.256) (0.162) (0.312) (0.243) (0.233) (0.152) (0.311) (0.283)
Medium 0.461* 0.015 0.296 0.371** 0.336 -0.170 0.262 0.192 0.515* 0.474*
(0.248) (0.286) (0.253) (0.162) (0.294) (0.259) (0.232) (0.154) (0.287) (0.258)
State 0.427* -0.032 0.045 0.118 -0.356 -0.126 -0.801*** -0.438*** 0.333 0.214
(0.244) (0.249) (0.240) (0.151) (0.287) (0.237) (0.247) (0.152) (0.277) (0.256)
Foreign -0.573* -0.222 -0.640*** -0.456*** -0.105 -0.198 -0.423** -0.212 -0.263 -0.144
(0.341) (0.235) (0.249) (0.167) (0.393) (0.211) (0.203) (0.145) (0.407) (0.351)
Sales change -0.002 -0.004** -0.004** -0.004*** 0.001 0.001 0.000 -0.001 -0.002 0.001
(0.002) (0.002) (0.002) (0.001) (0.001) (0.001) (0.001) (0.001) (0.002) (0.001)
Profitability1 -0.262 0.189
(0.253) (0.241)
Profitability2 -0.540*** -0.509***
(0.163) (0.153)
Account. Standards -0.111 -0.007 0.144 -0.113 0.094 -0.167 0.106 -0.020 0.443* 0.393*
(0.191) (0.164) (0.128) (0.103) (0.228) (0.218) (0.119) (0.098) (0.228) (0.211)
Year 1999 1.708*** 2.132***
(0.141) (0.139)
Year 2002 0.505*** 0.373***
(0.137) (0.125)
Constant 1.238 -1.534* -2.029*** -2.127*** 0.392 -1.868*** -2.978*** -2.446*** -2.713** -1.177
(1.116) (0.627) (0.592) (0.404) (1.011) (0.605) (0.575) (0.398) (1.175) (0.975)
Obs 696 988 1713 2722 687 1021 1728 2729 601 619
Log Likelihood -457.17 -521.55 -835.77 -1454.78 -305.79 -595.17 -969.11 -1581.58 -357.27 -391.82
LR chi2 50.52*** 72.43** 116.11*** 340.59*** 130.61*** 53.52*** 147.93*** 450.32*** 65.71*** 49.83***
Pseudo R2 0.05 0.07 0.065 0.1048 0.176 0.07 0.071 0.1246 0.09 0.07
Notes: ***, **, * denote significance at 1%, 5% and 10%, respectively; standard errors in parentheses. 
19
99
20
02
20
05
19
99
High Interest Rates Obstacle [Logit Model]
Hign collateral 
requirements and access 
to long-term loans as an 
obstacle to firm operation 
and growth
Financing Obstacle [Logit Model]
Po
ol
ed
C
ol
la
te
ra
l O
bs
ta
cl
e 
[L
og
it 
M
od
el
]
A
cc
es
s 
to
 L
on
g-
T
er
m
 L
oa
ns
 
O
bs
ta
cl
e 
[L
og
it 
M
od
el
]
20
02
20
05
Po
ol
ed
ii. High interest rates as an obstacle 
Firms may have access to external finance, but this does not imply that they are not pay-
ing a risk premium associated with asymmetrical information and other credit market 
frictions. Using a similar approach, in this section the focus is on another dimension 
of the financing obstacle, namely the degree to which firms perceive the high interest 
rates as an obstacle to growth and operation of their businesses. Results are presented 
in Table 4 (specifications 5–8). As can be seen, age has an estimated positive sign and is 
statistically significant in two specifications.18 The coefficient on small firms is statisti-
cally significant and positive in all the specifications, except in specification 6, and the 
coefficient on medium-sized firms is not significant. This indicates that, apart from the 
general financing obstacle, belonging to the group of small firms increases the likelihood 
of reporting higher obstacles for operation and growth due to high interest rates. Being 
state-owned firm has an estimated negative and statistically significant effect on perceiv-
ing high interest rates as an obstacle, except in two specifications. An explanation may be 
that either SOEs receive cheaper credit, or if they are credit rationed do not perceive high 
interest rates as an obstacle. As shown in the previous sections, the evidence suggests 
that SOEs are more likely to be credit rationed. The dummy variable for foreign firms is 
negative and statistically significant in the 2005 sample, giving some evidence that for-
eign firms may have access to cheaper loans than their domestic counterparts. The other 
explanatory variables, sales change and accounting standards are not significant. As in 
the previous model, the estimated coefficient on profitability is negative and statistically 
significant at the one percent level in 2005, but it is insignificant in 2002. In the pooled 
sample, the year dummies are positive and statistically significant, suggesting that in the 
earlier years firms perceived high interest rates as a greater obstacle for their operation 
and growth compared to 2005, broadly in line with the downward trend in lending rates 
in the SEE. 
 
iii. Collateral requirements and access to long-term loans as an obstacle 
In this section the determinants of firms’ self-reported obstacles for their operation and 
growth due to high collateral requirements and access to long-term loans are explored 
and the data is available for 1999 sample only. The estimates do not indicate a significant 
effect of age on firms perceiving collateral requirements and access to long-term loans as 
an impediment to their operation and growth (Table 4, specification 9 and 10). Small and 
medium-sized firms report significantly higher obstacles compared to the base group of 
large firms. Neither SOEs or foreign-owned firms are significantly different from private 
domestic firms in reporting these obstacles, although in all the specifications the state 
dummy is positive and foreign dummy is negative. The coefficient on sales growth is 
insignificant in all specifications. The estimated coefficient on accounting standards is 
positive and significant at the ten percent level only, suggesting that firms that use an 
accounting standard are more likely to perceive access to short- and long-term loans as a 
18 As previously, the specification replacing age with dummy variable taking value of one if the firm is less 
than five years old and zero otherwise is estimated. The newly defined age variable is not significant in either 
specification.
I. HASHI, V. Z. TOÇI  |  FINANCING CONSTRAINTS, CREDIT RATIONING AND FINANCING... 51
ECONOMIC AND BUSINESS REVIEW  |  VOL. 12  |  No. 1  |  201052
major obstacle compared to firms that do not use any standards, i.e., consider themselves 
as better quality borrowers.
To summarize, in terms of firms’ self-reported financing obstacles, size appears to mat-
ter as a determinant of various dimensions of finance as an obstacle to firms’ operation 
and growth. Small firms are more likely to face higher financing obstacles than large 
firms, while the evidence for medium-sized firms is weaker. The ownership effect also is 
an important determinant in various dimensions of financing obstacles. The regression 
results provide some evidence that foreign-owned firms are likely to face lower obstacles 
compared to their domestic counterparts. Evidence is not found that state-owned firms 
are likely to face lower financing obstacles compared to private companies, indicating 
that a hardening of budget constraints may have been put in place in the SEE. In general, 
firms with better prospects and good performance seem to face less financing obstacles. 
The firm’s age and whether the company uses accounting standards do not seem to have 
an impact on firms reporting lower financing obstacles, being insignificant in most of 
the regressions. 
3.3.4 Quantifying the effects: some examples
For all models discussed above, it is possible to quantify the probabilities of financing 
constraints or financial obstacles for different subgroups of firms. In this section we pro-
vide example of the predicted probabilities for two of these models. First, using the model 
with pooled data (Table 4, specification 4), we demonstrate the effect of size, ownership 
and time on the likelihood of the firm reporting finance as a major obstacle. Second, 
using the model of discouraged borrowers (Table 3, specification 3), we show the impact 
of profitability and the use of accounting standards on the likelihood of the firm being a 
discouraged borrower. The results, presented in Table 5, show that over time, in all firm 
categories (by size and ownership) there is a substantial decrease in the probability of 
reporting finance as a major obstacle. 
Table 5: Credit rationing of firms in SEE: quantifying the effects
Small 0.585 0.297 0.203 0.329 0.490 0.435 0.602
Medium 0.523 0.248 0.166 0.150 0.257 0.217 0.352
Large 0.431 0.185 0.121 0.080 0.145 0.120 0.210
State 0.460 0.204 0.134 0.145 0.249 0.210 0.343
Private Domestic 0.431 0.185 0.121 0.080 0.145 0.120 0.210
Foreign 0.324 0.126 0.080 0.067 0.124 0.102 0.182
Note: Continuous variables, age and sales change, are at their sample means; other dummies are zero; Profitability=1 is for profitable firms 
and 0 otherwise; Accounting Standards=1 is for firms that use national or international accounting standards and 0 otherwise.  
Predicted probabilities of reporting finance 
as a major obstacle
Predicted probabilities of credit rationing: the model of 
'discouraged' borrowers
1999 2002 2005
Profitability=1; 
Accounting 
Standards=1
Profitability=0; 
Accounting 
Standards=1
Profitability=1; 
Accounting 
Standards=0
Profitability=0; 
Accounting 
Standards=0
The table also shows that small firms consistently have a higher probability of reporting 
finance as a major obstacle compared to medium-sized and large firms. Medium-sized 
firms, too, have a higher probability of reporting finance as a major obstacle compared to 
large firms. The gap between the small and larger firms is decreasing over time suggest-
ing that the financing constraints for small firms are reducing in the SEE. It seems that 
there has been a shift towards lending to the small business sector in the region though, 
given the magnitude of the probabilities, finance remains problematic for growth and 
operation of a substantial proportion of firms, particularly small ones. In terms of the 
ownership effect, foreign-owned firms have consistently lower probability of reporting 
finance as a major obstacle compared to SOEs and their domestic counterparts.
Table 5 presents the predicted probabilities of firms being credit rationed by size and 
ownership, and considering the effect of profitability and following accounting stand-
ards. These two factors noticeably decrease the probability of credit rationing among all 
firm categories suggesting that banks screen their borrowers and increasingly are able to 
distinguish ‘good‘ from ‘bad‘ firms. However, as in the previous example, the probability 
of credit rationing decreases with the size of the firm – large firms consistently have a 
lower probability of being credit rationed compared to their smaller counterparts. This 
implies that a substantial proportion of small firms with ‘good‘ characteristics do not 
participate in the credit market either because they are rejected, or do not apply for a loan. 
4. CONCLUSIONS AND POLICY IMPLICATIONS
This paper assesses the significance and determinants of financing constraints, credit 
rationing and financing obstacles faced by firms in the SEE region by using a wide range 
of indictors and a large data set from the three rounds of BEEPS. These indicators include 
I. HASHI, V. Z. TOÇI  |  FINANCING CONSTRAINTS, CREDIT RATIONING AND FINANCING... 53
ECONOMIC AND BUSINESS REVIEW  |  VOL. 12  |  No. 1  |  201054
the proportion of investment expenditure financed by internal funds and bank loans, the 
probability of applying for a loan and being rejected, the probability of being a discour-
aged borrower and the importance of finance as an obstacle to the growth of firm. The 
empirical results suggest that the cohort of small firms is relatively more constrained. 
Compared to the group of larger firms, small firms rely more on internal funds and less 
on bank loans to finance investment, are less likely to apply for a loan, are more likely 
to be refused a loan, are more likely to fall within the group of discouraged borrowers, 
and face greater difficulties in accessing both short-and long-term loans. Furthermore, 
finance, high interest rates and high collateral requirements pose a greater obstacle to the 
operation and growth of small firms than larger firms. 
In terms of ownership, the evidence indicates that foreign-owned firms face lower fi-
nancing constraints compared to their domestic counterparts and SOEs. This may be 
explained by their ability to access financial sources from their parent companies and 
banks in their home countries. The estimates suggest that they do not rely much on the 
domestic banking sector – presumably because of better financing conditions in their 
home countries. State-owned firms, on the other hand, seem to face credit constraints 
from the financial sector, suggesting some hardening of budget constraints in the SEE 
region, at least in the bank-firm relationship. 
There was some support for the ‘bank screening hypothesis‘. It was found that more 
profitable firms, those with better prospects and firms that implement some accounting 
standards face fewer constraints, implying that banks do engage in screening their ap-
plicants to reduce information asymmetries and are able, to some extent, to distinguish 
‘good‘ from ‘bad‘ borrowers. However, the bank screening hypothesis is not the only ex-
planation. To the extent that screening is limited to observable characteristics of firms 
that make up each risk class, the  the unobserved heterogeneity remains within each class 
and credit rationing may still continue. There is also some evidence of self-selection in 
the credit market since firms with poor prospects do not enter the market (i.e., the likeli-
hood of adverse selection is less than expected). The evidence also suggests that there has 
been a substantial decrease in financing constraints in all firm categories. Importantly, 
the decrease of financing constraints is greater for small firms than for larger ones, in-
dicating a shift in the lending technology of banks towards small business sector. This 
may be a natural evolution, given the dominance of this sector in the countries under 
investigation. 
There is much room for government policies aimed at improving the position of small 
firms in the credit market. While loan guarantee schemes and other forms of subsidies 
to small firms may relax some of the financial constraints, these schemes also generate 
moral hazard and weaken the banks’ incentive to screen the applicants effectively (Vogel 
and Adams, 1997). Given the importance of collateral as a major element of financing 
constraint, there may be some room for the government to improve the operation of the 
collateral system and, as argued by DeSoto (2000), activate the large volume of ‘dead’ 
capital in these countries. The collateral system can be improved by strengthening the 
protection of property rights, improving the functioning of courts and rule of law, and 
establishing pledge registers for movable assets and land registers for immovable assets, 
formalising land and other property titles that need to be integrated in a unified system 
of registration to facilitate the quick verification of property rights at low cost. 
Credit rationing and financing constraints emanate from information asymmetries in 
the financial market. Government policies should aim to reduce the information asym-
metry problem by devising regulations on financial reporting and disclosure and the use 
of appropriate accounting and auditing standards. Screening and monitoring by banks 
will be easier when firms become more transparent and the accounting information be-
comes more reliable and meaningful, enabling banks to adopt lending technologies based 
on ‘hard‘ information. To the extent that weak transparency is an intrinsic feature of 
small firms, enhancing the availability of information through credit registers and other 
systems of notice will decrease the costs of screening loan applications.
REFERENCES:
Aghion, P. & Bolton, P. (1997). A theory of trickle-down growth and development. Review of Economic Studies, 
64, 151-172.
Anderson, R. & Kegels, C. (1998). Transition Banking: Financial Development of Central and Eastern Europe. 
London: Oxford University Press. 
Angelini, P. & Generale, A. (2005). Firm size distribution: do financial constraints explain it all? Evidence from 
survey data. Bank of Italy, Discussion Paper, No. 549.
Batra, G., Kaufmann, D. & Stone, A. (2002). The firms speak: what the world business environment survey tells 
us about constraints on private sector development, (mimeo).
Beck, T. et al. (2006). The determinants of financing obstacles. Journal of International Money and Finance, 
25, 932-952.
Besanko, D. & Thakor, A. (1987). Collateral and rationing: sorting equilibria in monopolistic and competitive 
credit market. International Economic Review, 28, 671-689.
Bigsten, A. et al. (2000). Credit constraints in manufacturing enterprises in Africa. ISA, Working Paper, 24. 
Bratkowski, A., Grosfeld, I. & Rostowski, J. (2000). Investment and finance in de novo private firms: empirical 
results from the Czech Republic, Hungary and Poland. Economics of Transition, 8, 101-116.
Charnes, A., Cooper, W. & Sueyoshi, T. (1988), A goal programming/constrained regression review of the Bell 
System break-up. Management Science, 34, 1-26.
Cressy, R. (1996). Are business start-ups debt-rationed. Economic Journal, 106, 1253-1270.
DeSoto, H. (2000). The Mystery of Capital: Why Capitalism Triumphs in the West and Fails Everywhere Else. 
New York: Basic Books. 
European Bank for Reconstruction and Development (2006). Transition Report. London: EBRD. 
Fazzari, S., Hubbard, R. & Petersen, B. (1988). Financing constraints and corporate investment. Brooking Pa-
pers on Economic Activity, 78, 141-195. 
I. HASHI, V. Z. TOÇI  |  FINANCING CONSTRAINTS, CREDIT RATIONING AND FINANCING... 55
ECONOMIC AND BUSINESS REVIEW  |  VOL. 12  |  No. 1  |  201056
Gertler, M. (1988). Financial structure and aggregate economic activity: an overview. Journal of Money, Credit, 
and Banking, 20, 559-588. 
Ghosh, P., Mookherjee, D. & Ray, D. (2001). Credit Rationing in Developing Countries: An Overview of the 
Theory. In Mookherjee, D. & Ray, D. (Eds.), Readings in the Theory of Economic Development. London: Black-
well Publishing.
Gilchrist, S. &. Zakrajsek, E (1995). The Importance of Credit for Macroeconomic Activity: Identification 
Through Heterogeneity. In Peek, L. & Rosengren, E. (Eds.), Is Bank Lending Important for the Transmission of 
Monetary Policy. Boston, Ma.: Federal Reserve Bank of Boston.
Greene, W. (1998). Sample selection in credit-scoring models. Japan and the World Economy, 10, 299-316. 
Guiso, L. (1998). High-tech firms and credit rationing. Journal of Economic Behavior and Organization, 35, 
39-59.
Heckman, J. (1979). Sample selection bias as a specification error. Econometrica, 47, 153-161.
Hersch, P., Kemme, D. & Netter, J. (1997). Access to bank loans in a transition economy: the case of Hungary. 
Journal of Comparative Economics, 24, 79-89.
Inderst, R. & Muller, H. (2007). A lender-based theory of collateral. Journal of Financial Economics, 84, 826-
859.
Jaffee, D. & Stiglitz, J. (1990). Credit Rationing. In Friedman, B. & Hahn, F. (Eds.), Handbook of Monetary 
Economics, Vol. II. Amsterdam: North-Holland.
Jappelli, T. (1990). Who is credit constrained in the U.S. economy?. Quarterly Journal of Economics, 55, 219-
234.
Jensen, M. & Meckling, W. (1976). Theory of the firm: managerial behaviour, agency costs and ownership 
structure. Journal of Financial Economics, 3, 305-360.
Konings, J., Rizov, M. & Vandenbussche, H. (2002). Investment and credit constraints in transition economies: 
micro evidence from Poland, the Czech Republic, Bulgaria and Romania. LICOS Discussion Paper, No. 112.
Levenson, A. & Willard, K. (2000). Do firms get the financing they want? Measuring credit rationing experi-
enced by small business in the U.S.. Small Business Economics, 14, 83-94.
Milde, H. & Riley, J.G. (1988). Signalling in credit markets. Quarterly Journal of Economics, 103, 101-129. 
Myers, S. & Majluf, N. (1984). Corporate financing and investment decisions when firms have information that 
investors do not have. Journal of Financial Economics, 13, 187-221.
Parker, S. (2002). Do banks ration credit to new enterprises? And should governments intervene?. Scottish 
Journal of Political Economy, 49, 162-193.
Parnitzke, T. (2005). Credit scoring and the sample selection bias, institute of insurance economics. University 
of St. Gallen, (mimeo). 
Perez, S. (1998). Testing for credit rationing: an application of disequilibrium econometric. Journal of Macr-
oeconomics, 20, 721-739.  
Perotti, E. & L. Vesnaver, L. (2004). Enterprise finance and investment in listed Hungarian firms. Journal of 
Comparative Economics, 32, 73-87.
Perotti, E. & Carare, O. (1996). The evolution of bank credit quality in transition: theory and evidence from 
Romania. CERT, Discussion Paper, No. 97/2. 
Riley, J. (1987). Credit rationing: a further remark. American Economic Review, 77, 224-227.
Rizov, M. (2004). Credit constraints and profitability. Emerging Markets Finance and Trade, 40, 63-83.
Saito, K. & Villanueva, D. (1981). Transaction costs of credit to the small-scale sector in the Philippines. Eco-
nomic Development and Cultural Change, 29, 631-640.
Schwartz, A. (1981). Security interests and bankruptcy priorities: a review of current theories. Journal of Legal 
Studies, 10, 1-37. 
Stiglitz, J. & Weiss, A (1981). Credit rationing in markets with imperfect information. American Economic 
Review, 71, 393-410.
Storey, D. (1994). Understanding Small Business Sector. London: Routledge Thomson Learning.
Townsend, R. (1979). Optimal contracts and competitive markets with costly state verification. Journal of 
Economic Theory, 21, 265-293.
Vogel, R. & Adams, D. (1997). Old and new paradigms in development finance: should directed credit be resur-
rected?. CAER II, Discussion Paper, No.2.
Williamson, S. (1986). Costly monitoring, financial intermediation, and equilibrium credit rationing. Quar-
terly Journal of Economics, 102, 135-145. 
Williamson, S. (1987). Costly monitoring, loan contracts, and equilibrium credit rationing. Journal of Mon-
etary Economics, 18, 159-179.
Wooldridge, J. (2001). Econometric Analysis of Cross-Section and Panel Data. Cambridge, Ma.: The MIT Press.
I. HASHI, V. Z. TOÇI  |  FINANCING CONSTRAINTS, CREDIT RATIONING AND FINANCING... 57
Appendix 1: Definition of variables
V
ar
ia
bl
e
Y
ea
r i
n 
w
hi
ch
 d
at
a 
ar
e 
av
ai
la
bl
e
D
ep
en
de
nt
 v
ar
ia
bl
es
In
te
rn
al
 F
un
ds
Th
e 
pr
op
or
tio
n 
of
 th
e 
fir
m
's 
in
ve
stm
en
t t
ha
t h
as
 b
ee
n 
fin
an
ce
d 
th
ro
ug
h 
in
te
rn
al
 fu
nd
s o
ve
r t
he
 la
st 
12
 m
on
th
s (
0-
10
0%
)
19
99
, 2
00
2,
 2
00
5
Ba
nk
 L
oa
ns
Th
e 
pr
op
or
tio
n 
of
 th
e 
fir
m
's 
in
ve
stm
en
t t
ha
t h
as
 b
ee
n 
fin
an
ce
d 
th
ro
ug
h 
ba
nk
 lo
an
s o
ve
r t
he
 la
st 
12
 m
on
th
s (
0-
10
0%
)
19
99
, 2
00
2,
 2
00
5
D
iff
ic
ul
ty
 o
f a
cc
es
sin
g 
sh
or
t-t
er
m
 lo
an
s
20
02
D
iff
ic
ul
ty
 o
f a
cc
es
sin
g 
lo
ng
-te
rm
 lo
an
s
20
02
D
isc
ou
ra
ge
d 
bo
rro
w
er
s
20
05
A
pp
lic
an
ts
D
um
m
y 
va
ria
bl
e 
ta
ki
ng
 ta
ki
ng
 th
e 
va
lu
e 
of
 1
 if
 th
e 
fir
m
 h
as
 a
pp
lie
d 
fo
r a
 b
an
k 
lo
an
 a
nd
 0
 o
th
er
w
ise
20
05
Re
je
ct
ed
 b
or
ro
w
er
s
D
um
m
y 
va
ria
bl
e 
ta
ki
ng
 ta
ki
ng
 th
e 
va
lu
e 
of
 1
 if
 th
e 
fir
m
 w
as
 d
en
ie
d 
a 
lo
an
 a
nd
 0
 o
th
er
w
ise
20
05
Fi
na
nc
in
g 
ob
sta
cl
e
19
99
, 2
00
2,
 2
00
5
H
ig
h 
in
te
re
st 
ra
te
 o
bs
ta
cl
e
19
99
, 2
00
2,
 2
00
5
Co
lla
te
ra
l o
bs
ta
cl
e
19
99
A
cc
es
s t
o 
lo
ng
-te
rm
 lo
an
s o
bs
ta
cl
e
19
99
Ex
pl
an
at
or
y 
V
ar
ia
bl
es
A
ge
Lo
ga
rit
hm
 o
f a
ge
19
99
, 2
00
2,
 2
00
5
Sm
al
l
D
um
m
y 
va
ria
bl
e 
ta
ki
ng
 th
e 
va
lu
e 
of
 1
 fo
r t
he
 fi
rm
 e
m
pl
oy
in
g 
2-
49
 e
m
pl
oy
ee
s a
nd
 0
 o
th
er
w
ise
19
99
, 2
00
2,
 2
00
5
M
ed
iu
m
D
um
m
y 
va
ria
bl
e 
ta
ki
ng
 th
e 
va
lu
e 
of
 1
 fo
r t
he
 fi
rm
 e
m
pl
oy
in
g 
50
-2
49
 e
m
pl
oy
ee
s a
nd
 0
 o
th
er
w
ise
19
99
, 2
00
2,
 2
00
5
La
rg
e
D
um
m
y 
va
ria
bl
e 
ta
ki
ng
 th
e 
va
lu
e 
of
 1
 fo
r t
he
 fi
rm
 e
m
pl
oy
in
g 
m
or
e 
th
an
 2
50
 e
m
pl
oy
ee
s a
nd
 0
 o
th
er
w
ise
19
99
, 2
00
2,
 2
00
5
St
at
e
D
um
m
y 
va
ria
bl
e 
ta
ki
ng
 th
e 
va
lu
e 
of
 1
 fo
r t
he
 fi
rm
 in
 m
aj
or
ity
 st
at
e 
ow
ne
rs
hi
p 
(>
50
%
) a
nd
 0
 o
th
er
w
ise
19
99
, 2
00
2,
 2
00
5
Pr
iv
at
e 
do
m
es
tic
D
um
m
y 
va
ria
bl
e 
ta
ki
ng
 th
e 
va
lu
e 
of
 1
 fo
r t
he
 fi
rm
 in
 m
aj
or
ity
 p
riv
at
e 
do
m
es
tic
 o
w
ne
rs
hi
p 
(>
50
%
) a
nd
 0
 o
th
er
w
ise
19
99
, 2
00
2,
 2
00
5
Fo
re
ig
n
D
um
m
y 
va
ria
bl
e 
ta
ki
ng
 th
e 
va
lu
e 
of
 1
 fo
r t
he
 fi
rm
 in
 m
aj
or
ity
 fo
re
ig
n 
ow
ne
rs
hi
p 
(>
50
%
) a
nd
 0
 o
th
er
w
ise
19
99
, 2
00
2,
 2
00
5
Sa
le
s c
ha
ng
e
Pe
rc
en
ta
ge
 c
ha
ng
e 
ov
er
 th
e 
la
st 
3 
ye
ar
s
19
99
, 2
00
2,
 2
00
5
Pr
of
ita
bi
lit
y1
D
um
m
y 
va
ria
bl
e 
ta
ki
ng
 th
e 
va
lu
e 
of
 1
 if
 th
e 
fir
m
 h
ad
 p
os
iti
ve
 g
ro
ss
 p
ro
fit
 to
 to
ta
l s
al
es
 ra
tio
 in
 2
00
1 
an
d 
0 
ot
he
rw
ise
20
02
Pr
of
ita
bi
lit
y2
D
um
m
y 
va
ria
bl
e 
ta
ki
ng
 th
e 
va
lu
e 
of
 1
 if
 th
e 
fir
m
 h
ad
 re
in
ve
ste
d 
pr
of
its
 in
 2
00
4 
fro
m
 p
ro
fit
s i
n 
20
03
 a
nd
 0
 o
th
er
w
ise
20
05
A
cc
ou
nt
in
g 
sta
nd
ar
ds
D
um
m
y 
va
ria
bl
e 
ta
ki
ng
 th
e 
va
lu
e 
of
 1
 if
 th
e 
fir
m
 u
se
s n
at
io
na
l a
nd
/o
r i
nt
er
na
tio
na
l a
cc
ou
nt
in
g 
sta
nd
ar
ds
 a
nd
 0
 o
th
er
w
ise
19
99
, 2
00
2,
 2
00
5
Se
ct
or
D
um
m
y 
va
ria
bl
es
 fo
r M
in
in
g,
 C
on
str
uc
tio
n,
 M
an
uf
ac
tu
rin
g,
 T
ra
ns
po
rt,
 T
ra
de
, R
ea
l E
sta
te
, H
ot
el
s a
nd
 R
es
ta
ur
an
ts,
 O
th
er
 
19
99
, 2
00
2,
 2
00
5
Co
un
try
D
um
m
y 
va
ria
bl
es
 fo
r A
lb
an
ia
, B
os
ni
a 
an
d 
H
er
ze
go
vi
na
, B
ul
ga
ria
, C
ro
at
ia
, M
ac
ed
on
ia
, R
om
an
ia
 a
nd
 S
er
bi
a 
an
d 
M
on
te
ne
gr
o 
19
99
, 2
00
2,
 2
00
5
Is
 th
e 
re
sp
on
se
 to
 th
e 
qu
es
tio
n:
 'c
an
 y
ou
 te
ll 
ho
w
 p
ro
bl
em
at
ic
 is
 L
ac
k 
of
 a
cc
es
s t
o 
lo
ng
-te
rm
 b
an
k 
lo
an
s 
fo
r t
he
 o
pe
ra
tio
n 
an
d 
gr
ow
th
 o
f y
ou
r b
us
in
es
s?
' (
w
ith
 1
 in
di
ca
tin
g 
no
 
ob
sta
cl
e;
 2
 a
 m
in
or
 o
bs
ta
cl
e;
 3
 a
 m
od
er
at
e 
ob
sta
cl
e 
an
d 
4,
 a
 m
aj
or
 o
bs
ta
cl
e)
Is
 th
e 
re
sp
on
se
 to
 th
e 
qu
es
tio
n:
 'c
an
 y
ou
 te
ll 
ho
w
 p
ro
bl
em
at
ic
 a
re
 C
ol
la
te
ra
l r
eq
ui
re
m
en
ts 
fro
m
 b
an
ks
 f
or
 th
e 
op
er
at
io
n 
an
d 
gr
ow
th
 o
f y
ou
r b
us
in
es
s?
' (
w
ith
 1
 in
di
ca
tin
g 
no
 o
bs
ta
cl
e;
 
2 
a 
m
in
or
 o
bs
ta
cl
e;
 3
 a
 m
od
er
at
e 
ob
sta
cl
e;
 a
nd
 4
, a
 m
aj
or
 o
bs
ta
cl
e)
D
ef
in
iti
on
Is
 th
e 
re
sp
on
se
 to
 th
e 
qu
es
tio
n:
 'c
an
 y
ou
 te
ll 
ho
w
 e
as
y 
w
ou
ld
 it
 b
e 
fo
r y
ou
r f
irm
 to
 o
bt
ai
n 
sh
or
t-t
er
m
 w
or
ki
ng
 c
ap
ita
l l
oa
n?
' (
w
ith
 1
 in
di
ca
tin
g 
im
m
po
ss
ib
le
; 2
, v
er
y 
di
ffi
cu
lt;
 3
, f
ai
rly
 
di
ffi
cu
lt;
 4
, f
ai
rly
 e
as
y 
an
d 
5,
 v
er
y 
ea
sy
)
Is
 th
e 
re
sp
on
se
 to
 th
e 
qu
es
tio
n:
 'c
an
 y
ou
 te
ll 
ho
w
 e
as
y 
w
ou
ld
 it
 b
e 
fo
r y
ou
r f
irm
 to
 o
bt
ai
n 
lo
ng
-te
rm
 b
an
k 
lo
an
 fo
r n
ew
 in
ve
stm
en
t?
' (
w
ith
 1
 in
di
ca
tin
g 
im
m
po
ss
ib
le
; 2
, v
er
y 
di
ffi
cu
lt;
 
3,
 fa
irl
y 
di
ffi
cu
lt;
 4
, f
ai
rly
 e
as
y 
an
d 
5,
 v
er
y 
ea
sy
)
Is
 th
e 
re
sp
on
se
 to
 th
e 
qu
es
tio
n:
 'c
an
 y
ou
 te
ll 
ho
w
 p
ro
bl
em
at
ic
 a
re
 H
ig
h 
In
te
re
st 
Ra
te
s 
fo
r t
he
 o
pe
ra
tio
n 
an
d 
gr
ow
th
 o
f y
ou
r b
us
in
es
s?
' (
w
ith
 1
 in
di
ca
tin
g 
no
 o
bs
ta
cl
e;
 2
 a
 m
in
or
 
ob
sta
cl
e;
 3
 a
 m
od
er
at
e 
ob
sta
cl
e;
 a
nd
 4
, a
 m
aj
or
 o
bs
ta
cl
e)
D
um
m
y 
va
ria
bl
e 
ta
ki
ng
 ta
ki
ng
 th
e 
va
lu
e 
of
 1
 if
 th
e 
fir
m
 w
as
 d
en
ie
d 
a 
lo
an
, a
nd
 a
m
on
g 
no
n-
ap
pl
ic
an
ts:
 fi
rm
s t
ha
t d
id
 n
ot
 th
in
k 
th
e 
ap
pl
ic
at
io
n 
w
ill
 b
e 
ap
pr
ov
ed
, c
ol
la
te
ra
l w
as
 to
o 
hi
gh
, p
ro
ce
du
re
s o
f a
pp
lic
at
io
n 
co
m
pl
ic
at
ed
, i
nf
or
m
al
 p
ay
m
en
ts 
re
qu
ire
d,
 se
ve
ra
l f
ro
m
 a
bo
ve
; a
nd
 0
 o
th
er
w
ise
 
Is
 th
e 
re
sp
on
se
 to
 th
e 
qu
es
tio
n:
 'c
an
 y
ou
 te
ll 
ho
w
 p
ro
bl
em
at
ic
 is
 F
in
an
ci
ng
 fo
r t
he
 o
pe
ra
tio
n 
an
d 
gr
ow
th
 o
f y
ou
r b
us
in
es
s?
' (
w
ith
 1
 in
di
ca
tin
g 
no
 o
bs
ta
cl
e;
 2
 a
 m
in
or
 o
bs
ta
cl
e;
 3
 a
 
m
od
er
at
e 
ob
sta
cl
e;
 a
nd
 4
, a
 m
aj
or
 o
bs
ta
cl
e)
ECONOMIC AND BUSINESS REVIEW  |  VOL. 12  |  No. 1  |  201058
Appendix 2: Various dimensions of financing obstacles, the ordered logit models
(1
)
(2
)
(3
)
(4
)
(5
)
(6
)
(7
)
(8
)
(9
)
(1
0)
(1
1)
(1
2)
A
ge
-0
.0
36
0.
14
5
-0
.1
87
0.
37
3*
*
0.
11
7
0.
22
9*
*
0.
04
1
0.
17
6
0.
22
3
0.
37
0*
**
-0
.0
72
-0
.1
07
(0
.1
96
)
(0
.2
00
)
(0
.1
93
)
(0
.1
90
)
(0
.1
66
)
(0
.1
10
)
(0
.2
22
)
(0
.1
99
)
(0
.1
59
)
(0
.1
14
)
(0
.2
00
)
(0
.2
21
)
Sm
al
l
0.
94
4*
**
1.
29
7*
**
0.
81
0*
**
0.
38
4*
0.
32
6*
*
0.
41
0*
**
0.
46
5
0.
21
9
0.
37
6*
*
0.
31
7*
**
1.
04
4*
**
1.
02
6*
**
(0
.2
08
)
(0
.2
14
)
(0
.2
57
)
(0
.2
04
)
(0
.1
59
)
(0
.1
23
)
(0
.3
05
)
(0
.2
01
)
(0
.1
61
)
(0
.1
23
)
(0
.2
80
)
(0
.2
66
)
M
ed
iu
m
0.
30
9
0.
46
1*
*
0.
65
5*
**
-0
.0
11
0.
20
0
0.
35
9*
**
0.
23
5
-0
.0
18
0.
28
0*
0.
26
5*
*
0.
45
9*
*
0.
49
1*
*
(0
.2
09
)
(0
.2
06
)
(0
.2
26
)
(0
.2
04
)
(0
.1
63
)
(0
.1
21
)
(0
.2
68
)
(0
.1
92
)
(0
.1
99
)
(0
.1
16
)
(0
.2
28
)
(0
.2
30
)
St
at
e
0.
22
8
0.
16
5
0.
45
4*
*
-0
.3
11
-0
.1
31
-0
.0
68
-0
.3
53
-0
.3
92
*
-0
.6
64
**
*
-0
.5
33
**
*
0.
15
2
0.
18
5
(0
.2
12
)
(0
.2
17
)
(0
.2
32
)
(0
.2
09
)
(0
.1
83
)
(0
.1
25
)
(0
.2
77
)
(0
.2
07
)
(0
.1
69
)
(0
.1
24
)
(0
.2
43
)
(0
.2
43
)
Fo
re
ig
n
-0
.4
86
**
*
-0
.5
46
**
*
-0
.6
26
**
-0
.4
05
**
-0
.5
57
**
*
-0
.4
83
**
*
0.
07
6
-0
.2
91
*
-0
.5
77
**
*
-0
.3
06
**
*
-0
.1
28
-0
.2
66
(0
.1
84
)
(0
.1
83
)
(0
.2
71
)
(0
.1
74
)
(0
.1
51
)
(0
.1
17
)
(0
.3
26
)
(0
.1
65
)
(0
.1
61
)
(0
.1
13
)
(0
.2
76
)
(0
.3
22
)
Sa
le
s 
ch
an
ge
 
-0
.0
05
**
*
-0
.0
04
**
*
-0
.0
02
-0
.0
03
**
-0
.0
02
*
-0
.0
03
**
*
0.
00
1
-0
.0
01
-0
.0
01
-0
.0
01
*
0.
00
1
0.
00
1
(0
.0
01
)
(0
.0
01
)
(0
.0
02
)
(0
.0
01
)
(0
.0
01
)
(0
.0
00
)
(0
.0
02
)
(0
.0
01
)
(0
.0
01
)
(0
.0
00
)
(0
.0
01
)
(0
.0
02
)
Pr
of
ita
bi
lit
y1
-0
.7
58
**
*
-0
.4
93
**
0.
06
7
0.
20
2
(0
.2
02
)
(0
.2
08
)
(0
.2
14
)
(0
.1
93
)
Pr
of
ita
bi
lit
y2
-0
.5
17
**
*
-0
.4
31
**
*
(0
.1
27
)
(0
.1
29
)
A
cc
ou
nt
in
g 
St
an
da
rd
s
0.
06
1
0.
07
6
-0
.0
13
-0
.0
15
0.
10
5
-0
.0
78
0.
07
8
-0
.1
28
0.
01
3
0.
00
8
0.
51
5*
**
0.
35
3*
(0
.1
35
)
(0
.1
33
)
(0
.1
77
)
(0
.1
23
)
(0
.0
89
)
(0
.0
77
)
(0
.2
28
)
(0
.1
23
)
(0
.0
56
)
(0
.0
08
)
(0
.1
89
)
(0
.2
02
)
Y
ea
r 1
99
9
1.
50
8*
**
1.
98
9*
**
(0
.1
10
)
(0
.1
23
)
Y
ea
r 2
00
2
0.
17
78
**
*
0.
36
1*
**
(0
.0
61
)
(0
.0
96
)
C
ut
1
-1
.2
99
**
-1
.1
47
**
-1
.4
67
**
0.
27
8
-0
.3
74
0.
32
0
-1
.9
41
**
-0
.2
47
-0
.0
90
0.
34
6
1.
04
6*
-0
.2
71
(0
.5
17
)
(0
.4
96
)
(0
.6
09
)
(0
.5
19
)
(0
.4
35
)
(0
.3
17
)
(0
.9
04
)
(0
.5
15
)
(0
.4
20
)
(0
.3
15
)
(0
.6
93
)
(0
.8
49
)
C
ut
2
0.
44
5
0.
46
3
-0
.8
15
1.
02
8*
*
0.
55
7
1.
14
8*
**
-1
.4
11
*
0.
82
7
0.
90
9*
*
1.
30
9*
**
1.
72
1*
**
0.
11
9
(0
.5
17
)
(0
.3
99
)
(0
.6
06
)
(0
.5
21
)
(0
.4
35
)
(0
.3
18
)
(0
.9
04
)
(0
.5
13
)
(0
.4
23
)
(0
.3
16
)
(0
.6
88
)
(0
.8
49
)
C
ut
3
1.
76
1*
**
1.
87
1*
**
0.
40
3
2.
15
7*
**
1.
78
9*
**
2.
34
9*
**
-0
.3
09
2.
04
1*
**
2.
15
0*
**
2.
44
6*
**
2.
72
1*
*
1.
04
1
(0
.5
19
)
(0
.5
00
)
(0
.6
01
)
(0
.5
28
)
(0
.4
36
)
(0
.3
20
)
(0
.9
01
)
(0
.5
15
)
(0
.4
25
)
(0
.3
18
)
(0
.6
93
)
(0
.8
48
)
C
ut
4
3.
74
2*
**
3.
54
9*
**
(0
.5
34
)
(0
.5
08
)
O
bs
96
3
94
7
69
6
98
8
17
13
27
22
61
3
10
21
17
28
27
29
60
1
61
9
Lo
g 
Li
ke
lih
oo
d
 -1
33
7.
76
 -1
36
1.
53
-8
06
.8
2
-1
29
0.
45
  -
22
70
.7
8
 -3
53
3.
00
-4
87
.7
5
-1
36
8.
5
-2
29
4.
57
-3
44
2.
07
 -7
53
.3
5
-6
37
.4
7
LR
 c
hi
2
22
0.
39
**
*
16
3.
92
**
*
68
.1
3*
**
95
.7
3*
**
15
6.
64
**
*
38
4.
05
**
*
13
4.
16
**
*
61
.7
**
*
16
7.
25
**
*
44
5.
93
**
*
92
.7
5*
**
58
.9
3*
**
Ps
eu
do
 R
2
0.
08
0.
06
0.
04
0.
04
0.
04
0.
05
0.
12
09
0.
04
0.
04
0.
07
0.
06
0.
05
N
ot
es
: *
**
, *
*,
 *
 d
en
ot
e 
si
gn
ifi
ca
nc
e 
at
 1
%
, 5
%
 a
nd
 1
0%
, r
es
pe
ct
iv
el
y;
 R
ob
us
t s
ta
nd
ar
d 
er
ro
rs
 in
 p
ar
en
th
es
es
. 
C
ol
la
te
ra
l 
O
bs
ta
cl
e 
A
cc
es
s 
to
 L
on
g-
Te
rm
 L
oa
ns
 
O
bs
ta
cl
e 
   
 
H
ig
h 
In
te
re
st
 R
at
e 
O
bs
ta
cl
e 
D
iff
ic
ul
ty
 
A
cc
es
si
ng
 S
ho
rt-
Te
rm
 lo
an
s 
D
iff
ic
ul
ty
 
A
cc
es
si
ng
 L
on
g-
Te
rm
 lo
an
s 
Fi
na
nc
in
g 
O
bs
ta
cl
e 
I. HASHI, V. Z. TOÇI  |  FINANCING CONSTRAINTS, CREDIT RATIONING AND FINANCING... 59
ECONOMIC AND BUSINESS REVIEW  |  VOL. 12  |  No. 1  |  201060