The relationship between liquidity risk and credit risk in Islamic banking industry of Iran

Article history: Received October 28, 2012 Received in revised format 18 February 2013 Accepted 19 February 2013 Available online February 27 2013 An integrated risk management is a process, which enables banks to measure and manage all risks, simultaneously. The recent turbulent chaos on banking industry has increase the relative importance of risk management, more than before. This paper investigates the relationship between credit risk and liquidity risk among Iranian banks. The proposed study includes all private and governmental banks as population over the period 2005-2012. The results Pearson correlation has disclosed a positive and meaningful relationship between credit and liquidity risks. Bank size also impacts on two mentioned risk factors but we there seems to be no relationship between financial chaos and type of ownership with risk factors. © 2013 Growing Science Ltd. All rights reserved.


Introduction
During the past two decades, there have been significant changes on banking industry in the world due to financial crisis in this sector in 2008 (Lando, 2009;Fiordelisi et al., 2010).Many banking officials attempt to put more restrictions on giving loan to business owners in an attempt to prevent any trouble making issue.There are also various studies on relationship between different banks' characteristics such as size, market capitalization, etc. (Wong et al., 2008).Salas and Saurina (2002), for instance, investigated credit risk in two institutional regimes by studying two Spanish commercial and savings banks.Dičevska (2012) performed an investigation on credit risk and established a system for credit risk management in changing economic conditions in Macedonian banks.Cifuentes et al. (2005) investigated liquidity risk in a system of interconnected financial institutions when these institutions were under regulatory solvency constraints and marked their assets to market.According to their survey, when the market's demand for illiquid assets was less than perfectly elastic, sales by distressed institutions depressed the market prices of such assets.They studied the theoretical basis for contagious failures, quantified them through simulation exercises and reported that liquidity requirements on institutions could be as efficient as capital requirements in forestalling contagious failures.Michalak and Uhde (2012) provided some empirical evidence that credit risk securitization had a negative effect on the issuing banks' financial soundness.For this purpose, they used a unique sample of 749 cash and synthetic securitization transactions issued by 60 stock-listed bank holdings in the EU-13 plus Switzerland over the period over the period [1997][1998][1999][2000][2001][2002][2003][2004][2005][2006][2007].They reported a negative influence of securitization on bank profitability and capital environment as well as a positive relationship between securitization and the issuing bank's return volatility.They underlined that the decision by the Basel Committee to enhance the new Basel III framework in the field of securitization was a step in the right direction.

The proposed study
The proposed study of this paper considers the effects of three variables, namely ownership type, bank size and financial crises on two risk components including credit and liquidity risks.Fig. 1. shows details of our proposed model.Any positive value for liquidity risk (LR) indicates that bank cannot guarantee incidents.

Berger-Bouwman (BB) measure
Berger-Bouwman (BB) factor is calculated as Cat Fat/Total Assets while credit risk (CR) is measured as follows, ln( / ) , where ROA is return on assets, CAP represents the ratio of total equities on total assets and σ ROA is the standard deviation of ROA.Obviously, the higher value represents the higher risk.Credit risk (CR) ratio can also be calculated as follows, In this study, bank ownership is a dummy variable, which is equal to zero for governmental banks and one for private banks.Bank size is also calculated by taking the log of total assets and Financial crises is also a dummy variable, which is equal to zero when there is no crises and one during the financial turbulence.In our study, there were 144 observations and 45.8% of the data were associated with governmental banks while 54.2% of the observations were associated with private banks.In addition, 37.5% of the observations belong to before crisis and 62.5% of the data were associated with after crisis.Table 1 demonstrates some basic statistics on bank size.The results of Table 2 indicate that data do not seem to be normally distributed.Table 3 shows similar results for CR variable.The results of Table 3 for CR indicate that the results are away from normal distribution.Table 4 presents the same descriptive results for variable LR and we could make similar conclusion that LR was not normally distributed.In order to have a better understanding on the nature of data, we use Kolmogorov-Smirnov and Shapiro-Wilk tests and Table 5 demonstrates the results of our experiment for 144 observations.As we can observe from the results of Table 5, only bank size is normally distributed and other variables are not normally distributed when the level of significance is five percent.

The relationship between credit and liquidity risks
In this section, we present details of our findings for the relationships between different variables based on Pearson correlation ratios.Table 6 demonstrates the results of our survey.In this study credit risk is measured based on two attributes of CR and CRz-score.In addition, liquidity risk is calculated based on two attributes of LR and LR BB.Based on the results of Table 6, we can observe that there is a meaning and reverse relationship between CRz-score and LR when the level of significance is five percent.There are also meaningful and reverse relationships between CRz-score and LR, between CRz-score and LR, when the level of significance is five percent.However, the relationship between CR and LR and between CR and LR BB are meaningful but positive when the level of significance is five percent.

2.2.The results of Pearson correlation with bank size as control variable
We have accomplished the same results as explained in previous section when there is an additional variable, bank size, and Table 7 demonstrates the results of our survey.The results of Table 7 show that in the presence of bank size, the relationship between CRz-score and LR is reduced from -0.194 to -0.062 and although the relationship is still negative but it is not statistically significance.The same result holds for the relationship between CRz-score and LR BB and we observe that the relationship is reduced from -0.179 to -0.046 but it is not statistically significance.However, the positive relationship between CR and LR BB has been increase from 0.259 to 0.281 and it is still significant even when the level of significance is one percent.The results of Table 8 show that the relationship between CRz-score and LR is negative for governmental banks and it is statistically significance when the level of significance is five percent.In addition, the relationship between CRz-score and LR is negative for private banks but it is not statistically significance when the level of significance is five percent.There is also a negative and meaningful relationship between CRz-score and LR BB between governmental banks when the level of significance is one percent.Despite the fact that the same relationship holds for governmental banks, the relationship is not statistically significance.The relationship between CR and LR is positive in governmental and private banks but it is not statistically significance.Finally, in spite the fact that the relationship between CR and LR BB is not statistically significance for governmental banks, it is statistically significance for private banks.In order to compare the relative effect of various factors between governmental and private banks we have calculated Fisher correlation ratio based on the following

The results of Pearson correlation ratio for credit and liquidity risks between private and governmental banks
The result of Table 9, none of z value is not statistically significance leaving us to conclude that ownership type does not play an important role on different risk components.

Table10
The results of Pearson correlation ratios before and after crisis According to the results of Table 10, there is a reverse and meaningful relationship between CRzscore and LR before crisis when the level of significance is one percent but this relationship is not statistically significance after crisis.The relationship between CRz-score and LR BB is not significance either before or after the crisis.The relationship between CR and LR is positive and meaningful before crisis but it is not meaningful after crisis.The relationship between CR and LR BB is not statistically significance before crisis but it is statistically significance after crisis happens.
In order to compare the relative effect of various factors between governmental and private banks we have calculated Fisher correlation ratio based on the following The results of Table 11, none of z values is not statistically significance leaving us to conclude that crisis does not play an important role on different risk components.

The results
In this section, we summarize the results of our survey for four hypotheses of this survey.Next, we first present the results of our investigation on four various types of hypotheses.

The first hypothesis: The relationship between credit and liquidity risk components
The first hypothesis of this survey considers whether there is a positive and meaningful relationship between credit and liquidity risk.The following summarizes the results of our survey, As we can observe from the results, there is a meaningful relationship between these two variables, the negative sign is consistent with our hypothesis since a reduction in risk will increase the credit and we can confirm the first hypothesis.Similarly, we perform the following test between two variables LR BB and Z-score as follows, As we can observe, there are similar results and we can confirm there is a meaningful relationship between these two variables when the level of significance is five percent.In addition, the relationship between LR and CR as well as between LR BB and CT are meaningful when the level of significance is five percent and the results are summarized as follows,

The second hypothesis: The effect of ownership on the relationship between credit and liquidity risk
The results of investigating the effect of ownership on two risks of LR and CRz-score are examined as follows, ,


= 0.792 < 1.96 The results indicate that ownership does not have any impact on two risks of LR and CRz-score when the level of significance is five percent.Similar investigation has been performed between LR BB and CRz-score and the result is Zob =1.350 < 1.96 and this confirms that ownership does not have any impact on LR and CRz-score when the level of significance is five percent.The same conclusions hold for the effect of ownership for the relationships of LR and CR as well as LR BB and CR when the level of significance is five percent.In other words, ownership does not play important role on these risk components.

The third hypothesis: The effect of bank size on the relationship between credit and liquidity risk
The results of investigating the effect of bank size on two risks of LR and CRz-score are examined as follows, The result of Pearson correlation ratio does not indicate that bank size has any impact on the relationship between LR and CRz-score when the level of significance is five percent.Similarly, the results of investigating the effect of bank size on the relationship between LR BB and CRz-score is p ob =0.62>p oc =0.05, which means there is no meaningful relationship.However, the effect of bank size on relationship between CR and LR as well as CR and LR BB are equal to p ob =0.03<p oc =0.05 and p ob =0.002<p oc =0.05, which means the bank size influences these two pairs of risk factors, significantly.

The fourth hypothesis: The effect of crisis on the relationship between credit and liquidity risk
The results of investigating the effect of ownership on two risks of LR and CRz-score are examined as follows, , , The results indicate that crisis does not have any impact on two risks of LR and CRz-score when the level of significance is five percent.Similar investigation has been performed between LR BB and CRz-score and the result is Zob =-0.69 < 1.96 and this confirms that crisis does not have any impact on LR and CRz-score when the level of significance is five percent.The same conclusions hold for the effect of ownership for the relationships of LR and CR as well as LR BB and CR when the level of significance is five percent.In other words, crisis does not play important role on these risk components.

Other results
We have also performed regression analysis on relationship between the effects of CRz-scor, LR BB , LR and CR as independent variables and bank size, crisis and ownership as dependent variables.
Table 12 shows the results of testing the first regression model where LR is the dependent variable.As we can observe from the results of Table 12, while crisis has not significant impact on LR, ownership and bank size have positive and meaningful effects on LR.Similar results are executed on the same data where LR BB is the dependent variable and Table 13 demonstrates the results as follows,

Table 13
The results of regression analysis for the first model when LR BB is dependent variable (R-Square=0.052)The results of Table 13 indicate that only bank size maintains important effect and two other factors, ownership and crisis, do not play important role on LR BB.Another investigation is to consider CRzscore as dependent variable and Table 14 presents details of our findings,

Table 14
The results of regression analysis for the first model when CRz-score is dependent variable (R-Square=0.012)The results of Table 14 also indicate that none of the independent variables has any meaningful impact on CRz-score.Finally, we have investigated the impact of dependent variables on CR and found out that none of them had any meaningful impact on CR.

Conclusion
In this paper, we have performed an empirical investigation on measuring the effect of some Iranian banks on two credit risk factors.The proposed study has used Pearson correlation tests to investigate the relationships.We have also considered some linear regression models, where three variables of banks size, ownership and financial crisis are considered as independent variable and different risk factors were considered as dependent variable.The results of regression analysis have indicated that while crisis has not significant impact on LR, ownership and bank size had positive and meaningful effects on LR.In addition, only bank size maintained important effect on LR BB and two other factors, ownership and crisis, did not play important role on LR BB.Finally, we have investigated the impact of dependent variables on CR and found out that none of them had any meaningful impact on CR.

Fig. 1 .
Fig. 1.The proposed framework of the study

Table 1
Basic statistics associated with bank size

Table 4
Basic statistics associated with LR

Table 5
The results of Kolmogorov-Smirnov and Shapiro-Wilk tests

Table 6
The summary of statistical observations for the implementation of Pearson correlation ratio

Table 7
The results of Pearson correlation ratios

Table 8
The results of Pearson correlation ratios for private and governmental ownership

Table 12
The results of regression analysis for the first model when LR is dependent variable