Risk of Credit and Informational Asymmetry
In the case of financial distress, the banking decision is limited to the alternative between bankruptcy or financial support. Generally, when information is accessible, the bank chooses the decision which maximizes its profit. In addition, when information is private, the bank chooses the decision to bankruptcy sanction as the illiquid company. On the contrary, if the bank has an informational advantage compared to the other creditors, it is able to exploit it and in this case, the probability of the statement of affairs of an illiquid company is then reduced at least in the short term in the form of an abusive support.
We propose to check empirically the impact of the informational problems on the bankruptcy of the companies starting from the Belgian accountable data. A Probit model is used. The endogenous variable is the fact of having or not to file for bankruptcy, the explanatory variables include the revealing variables of the financial, economic and legal situation of the companies, as well as of the proxy variables of informational asymmetries. The assumption of the abusive support is not always confirmed. The results on the contrary appear that the decision to bankruptcy is all the more severe than information is private and hardly accessible. Key words: credit risk- Informational Asymmetry- Banking Decision-
Since the Eighties, several theories insisted on the fact that the risk of credit constitutes the first internal cause of banking failure. It is defined as "non performance" of the counterpart generating a probable loss on the level of the bank.
More precisely, it comes from the incapacity of the debtor to respect all or part of its obligations towards the creditor. This last will undergo this non-payment; it can determine the capacity to require to file the balance-sheet of the debtor. However, it may find it beneficial to give a financial support for its customer in hope of a further refunding.
In addition, the decision of credit put back on the effective treatment of information by the bank, which requires the need for an organisational structure adapted to the type of information. Indeed, it is in this context of the banking relation that the asymmetry of information between the parts can cause a problem of moral risk leading to an ineffective decision of credit. Thus, it is always necessary to follow control by mechanisms of adequate organization and governance banking. In addition, there are other causes being able to influence the decision of bank credit as institutional and legal environment, regulation and prudential supervision of the banks, as well as the structure and the discipline of market. The latter constitute pillars 1, 2 and 3 of the Basle II reform. These processes suppose that the bank holds capital funds in adequacy with its risk profile making it possible to improve the communication of information and to encourage the application of the beneficent banking practices.
Banking industry is strongly dependant on information. The bank is always confronted with the problem of asymmetry of information, and to solve it, it can acquire two types of information: Hard information, in external, by the means of public information (Rating, Score...) and of Soft information, in internal, by the means of the relation of customers. That supposes two methods of loans attribution: the bank with the act versus the bank of relation (Peterson, 2004)
It is well-known in the literature that the division of information and the respect of the international standards improve the conditions
of access to the credit by reducing the problems of adverse selection and moral risk. The banks must be organized to apprehend and treat the risk of credit, in particular international statutory constraints supporting on them (Basle II). The object of this article is to propose the principal theoretical determinants of the banking decision facing the non-payment of a debtor and to propose thereafter an empirical checking of their impact to bankruptcy. With this intention, we will use the Probit model to determine the probability to bankruptcy through in particular two samples of healthy and failing companies belonging to the Belgian sector for period 20012004 and through a series of indicators or variables related to the economic, financial and legal situation of the company and accountable variables able to reduce the existence of informational asymmetries. We are inspired partly by the empirical studies where the informational problems seem like exogenous variables, knowing that it is difficult to apprehend directly the informational problems; the studies consider variables which make it indirectly possible to give an account of it.
In the case of participation of several creditors in the company in addition to the bank, there is risk of appearance of the conflicts of interest and the probability of ‘to go into liquidation’ increases because the financial support falls on the bank (Bulow and Shoven, 1978).
If on the other hand a coalition between the shareholders and the principal bank of the company is suitable to be trained to the detriment of bond creditors and the probability of a support for the company would be the most strong.
1- THE DATA
Initially, we select the Belgian companies starting from the base provided by World scope Global, then, we present revealing accountable variables of the existence of the problems of informational asymmetries. According to some empirical studies, we suppose that these problems seem exogenous variables. Finally we describe, the explanatory variables selected to describe the economic, financial and legal situation of the firm.
The selected sample includes healthy companies and failing companies. All are firms of the sector of the intermediate goods subjugated to the corporation tax.
The decision of setting in bankruptcy of the company by the bank occurs at the time of the non-payment. Thus, we consider for the failing companies the last assessment available before the bankruptcy, the data are provided for 4 years (2001-2004).
The Year 2001 being best represented, it is that which we also retained for the non failing companies ( were not the subject of legal procedures).
We retain the assumption according to which the failing company does not respect its financial engagements i.e. which had the financial expenses higher than the gross surplus of exploitation.
50 failing companies and 120 non-failing companies were thus selected.
1.2 the choice of the variables:
The informational problems are apprehended with difficulty, but some variables indirectly make it possible to give of it an account and they are gathered in the context of reduction of informational asymmetries. - Variables able to reduce the existence of the informational problems:
To reduce the asymmetries existence of information, one of the first variables selected is the size of the firm or more precisely the turnover or the total numbers. Indeed, more the company is of significant size, more it is subjected to legal constraints in term of production of information. We thus retain the total numbers (NUM).
A second variable used is the duration of the relation passed with the lender or the age of the company. The relations of long term weaken asymmetries of information, reduce the costs of monitoring and generate thereafter a relaxation of the supervision of the company by the bank.
Asymmetries of information are also reduced either when the company is with dimensions because the latter will be obliged to produce reliable information or when the bank is
shareholder of the company because in this case it has access to the private information more easily than as a simple creditor.
This table checks off the various revealing variables of asymmetries of information:
The revealing variables of growth opportunity are the expenditure in research and development which represent a source of future growth. We thus retain the relationship between the expenditure in R&D and the turnover noted: (RD/ TUR)
Also, the intangible credits represent from every angle growth appropriateness. Indeed if the accountable assessment records them with their historical value, on the contrary the financial market, by its evaluation of the firm, accounts for the anticipated growth appropriatenesses. The productive investment which is at the origin of the growth opportunities; thus, we consider the ratio productive capital expenditure on total assets (PI/ASS) Some studies try to identify the risk of substitution of assets. It is all the more weak as the productive structure is rigid, therefore that the ratio tangible assets immobilized on total assets is high or that the ratio of the productive equipment on the engaged capital is significant, we retain the ratio (PEQ/K), indeed the risk of substitution of asset is all the more significant since the company carries out new investments. The ratio debts towards the group and of partners on total assets, (PAR/ASS) is used as proof of the confidence of partners within the context of informational asymmetries. Indeed more this ratio is high; more the risk of substitution of assets is weak, because the associates would be the first despoiled in the case of excessively risky investment.
2- Variables relating to the economic, financial and legal situation of the companies:
Other explanatory variables introduced into the model will be supposed to affect the banking decision. In the case of liquidation, the sold goods are the tangible permanent assets and stocks. So, we retain the market value of the firm as first explanatory, it will be apprehended by the ratio (immobilized tangible assets + total stocks) / total asset noted (LIQ). The crisis of illiquidity is measured by the margin, between the gross surplus of exploitation and the financial expenses normed by the noted turnover, (LIQ/ TUR) A third variable will be retained in our empirical model to realise of the clear situation of the company under consideration by the total assets ratio / total debt (ASS/DEB). The debt position is consisted of the banking loans, of the current banking contests, other financial debts and debts of exploitation; therefore, we retain the relationship between the whole of the short-term debt and the whole of financial debt considered by the banking ratio debt of shortterm/ total debt (shtD/DEB)
This table presents a synthesis of the used ratios. Explanatory variables that they represent and the economic phenomenon which they describe.
According to the descriptive static of these various exogenous variables, we could confirm the theoretical results. Indeed, through the two underpopulations, the failing companies are characterized on the average by a liquidity ratio (LIQ/TUR) and a lower net situation (ASS/DEBT) whereas by comparisons with the healthy companies, they are strongly financed by shortterm debts (shtD/DEBT)
For the variables able to reduce the informational problems, only averages of the total numbers (NUM) and ratios loan of partners / total assets (PAR/ASS) which are in conformity with the theory. Indeed, more the company is of significant size, more it is subjected to constraints in term of production of information.
In the same way for the ratio (PAR/ASS), more this ratio is high, plus the risk of substitution of assets is weak, because the associates would be the first stripped in the event of excessively risky investment.
The productive capital expenditure reported in the total assets (PI/ASS), are higher for the non-failing companies. Indeed they represent a signal of the hope of the future profit. However, the market value (LIQ) is not significantly different between the two sub-populations. 3- The methodology of research and estimates:
The methodology of research consists in testing two models: a basic model in perfect information and a total model by using variables able to realize the existence of asymmetries of information.
We seek to check if the existence of asymmetry of information has an influence on the fact that a company files for bankruptcy or not by
taking account of the two assumptions in the case of financial distress.
Indeed, the creditor who undergoes the nonpayment holds the capacity to require the liquidation of the debtor (assumption 1). But, it can have also interest to give a financial support in the hope of a later refunding (assumption2). According to this study. The dichotomous variable Y will be fixed and a model Probit is used.
By dichotomous model, we understand a statistical model in which the explained variable can take only two methods (dichotomous variable).
Then, it concerns generally to explain the unexpected arrival or the non-unexpected arrival of an event.
Among the most known applicability, we quote that which consists of the modelling of the default risks in a loan relation, or any other form of contract of service (telephone subscription contract, contract of assistance etc...). We consider for example a dichotomous variable taking two methods: "rupture of the contract" and "continuation of the contract ". They are the techniques of bases here, methods of scoring largely used in the banking sector and in the telecommunications sector. We pose, [l, N]:
Let us suppose that we have N observations y, whatever i = 1..., N: of a coded dichotomous endogenous variable yi = 1 or yi = 0 per convention, when in parallel the observations of K exogenous variables are:
In this case, the simple linear model is written: Where indicates a vector of K unknown parameters and where the disturbances.... are supposed to be independently distributed. Then, we can highlight several problems involved in the use of this simple linear specification to model our dichotomous variable. The variable yi is of qualitative type while the sum.
The dichotomous probit model admits for explained variable, the probability of appearance of this event conditionally to the exogenous variables. Thus, the following model is
Where the function F (.) Indicates a distribution function. In the case of this model, the function of distribution F (.) corresponds to the function of distribution of the reduced centred normal law
Therefore, for a given value of the vector of exogenous and vector of the parameters we can define the following model: For Yi is equal to 1 when a company i is failing and Yi is equal to 0 when company i is non-failing. X is the vector of the explanatory variables The Estimated Probit model is written:
The errors ?i follow a reduced centred normal law.
The vector will be estimated by the method of the Max of likelihood.
The quality of the adjustment of the model is tested by the report of likelihoods.
Where F expresses the function of distribution of the reduced centred law.
This test consists in comparing the
probability of the model considered noted L with noted likelihood of the simple model where the only explanatory variable considered is the constant. The null assumption HO corresponds to the nullity of the whole of the parameters. Under HO, the statistics S = 2(ln L -ln follow a law of chi-two of degree of freedom
K (a number of explanatory variables except constant, are the dimension of X minus 1). The significance of each estimated parameter βj (j=1,…, K-1) is also tested thanks to the report of likelihoods. The null assumption HO is the nullity of the parameter ?j. The likelihood of the estimated model is compared with the likelihood of the constrained model estimated under Under HO, the statistics follow chi-two of degree of freedom 1.
In the case of the unvaried dichotomous model, the construction of likelihood is extremely simple. Indeed, with the Yi event = 1 is associated the likelihood and in event Yi=0 corresponds the likelihood . . This makes it possible to regard the actual values Yi as the achievements of a binomial process with a likelihood of
By distinguishing the observations Yi = 1 those for which we have Yi = 0, the log-likelihood can be written in the form: The estimator of the maximum of likelihood of the parameters... is obtained by maximizing either the function of likelihood or the function of log likelihood log
4-results and conclusions:
Estimation of the model of reference (perfect information): From this model, we want to check that in the case of symmetry of information, the bank tends to support its company customer.
This table represents for each exogenous variable J the estimate of the parameter The number between brackets is the associated probability. This model is overall significant; each found sign confirms the theoretical results. The support of a company is thus strongly correlated for the quality of its economic and financial situation. This support which is generally abusive. However, such a conclusion is to be moderated: If the support is based on the reliability of the information held by the bank, then the accountable assessments easily available always do not account for the true health of the company? Estimation of the model 2(in asymmetry of information: introduction of the informational variables).
To introduce these variables within the model, we will carry out an ascending selection We carry out an ascending selection (the highest report of likelihood). We select initially the explanatory variable most correlated with the dependent variable. Then, we select, among those which remain, the explanatory variable whose partial correlation is highest (by keeping constant the variables already selected). And so on as long as there remain variables candidates whose partial coefficient of correlation is significant. We retained only the variables from which the coefficient is significantly different from zero with the threshold 10%. According to this assumption, one eliminated from the model the coefficients associated with the expenditure in research and development reported to the turnover and the share the productive equipment in F total assets since they are not significantly different from zero with the threshold of 0.1.
Only total number NUM, the productive capital expenditure brought back in the total assets.PI/ ASSand the share of the loans granted by the associates in the whole of the total assets.PAR/ASSare integrated into model 1 to constitute model 2.
Model 2 is written then:
From this model, we empirically will check the impact of each informational variable on the setting in bankruptcy. This table represents for each exogenous variable, the estimate of the parameter the number between brackets is the associated probability:
This tested model is significant. We want moreover test the assumption HO according to which the total model is not significantly more explanatory than the basic model.
Following the calculation of the report of likelihood S=2(lnL the likelihood of model 2 and L is the likelihood of model 1.
The coefficients of the exogenous variables relating to model 1 remain significant, with the awaited signs. The variables introduced into model 2 have a negative impact on the probability of setting in bankruptcy. This result confirms the idea that the setting in liquidation by the bank is influenced by the existence of the problems of asymmetries of information.
The productive investment reported in the total assets intervenes negatively in the probability of setting in liquidation. Indeed, the growth appropriatenesses encourage the bank to support the company.
According to these results, the decision of setting in bankruptcy can be softened only when asymmetries are reduced.
As a conclusion and through this methodology of research, two assumptions are taken into account: the assumption of the support of the company which is strongly correlated with the quality of its economic and financial situation, indeed, if private information would be held by the bank, all the accountable data are easily available, the bank can decide in the case of financial distress to support its customer in the hope of a later refunding. The assumption of setting in bankruptcy which is all the more severe as private information is difficult to reach, in this case, the bank in case of doubt and in fear of undergo the loss, the setting in automatic bankruptcy of its customer decides. The estimated model can be more relevant by integrating several elements:
More detail by creditor of the authorized credits, this will be able to confirm the theoretical results.
To increase the sample of the companies (healthy and failing) while varying the branches of industry this then supposes to integrate a new variable in the model which is the variable Sector.
To analyze the behaviour of the bank over several periods and long-term this could improve the explanation of the setting in bankruptcy of the company
Bédué, Lévy, 1997, " relation banqueentreprise et coût du crédit ", revue d'économie financièren°39-1997. Bell,J et D Pain,2000," leading indicator models of banking crises_ a critical review", financial stability review, banque d'angleterre, n°9, article3,pp,113-129, décembre. Berger, An and Udell, GF, 1992," Some evidence on the empirical significance of credit rationing," the journal of political economy,vol100,n°5, October,1047-1077. Berger et udell, 1995,"Relationship lending and lines of credit in small firm finance", journal of business, vol.68,n°3,pp351-377. Berger et udell, 2002," small business credit availability and relationship lending: the importance of bank organisationnal structure",economic journal,112,pp34-53. Berkson J. ( 1944), " Application of the Logistique Function to Bio-Assay ", JASA, 39, 357-365. Berkson J. (1951), "Why I prefer Logit to Probit", Biometrics, 7, 327-339. Berlin M., Loeys J. (1988), " Bond Covenants and Delegated Monitoring ", the Journal of Finance, vol. 43, p. 397-412. Berlin M, ET Mester L, 1992, " debt covenants and renegociation", journal of financial intermediation, vol.2, p95-133.