Research in International Business and Finance 45 (2018) 577–587

Contents lists available at ScienceDirect

Research in International Business and Finance

journal homepage: www.elsevier.com/locate/ribaf

Behavioral explanation for risk taking in Islamic and conventional

T
banks
⁎
Ines Ben Salah Mahdi , Mouna Boujelbène Abbes
Faculty of Economics and Management of Sfax, Laboratory URECA, University of Sfax, Street of airport, km 4.5, LP 1088, Sfax, 3018, Tunisia

AR TI CLE I NF O AB S T R A CT

Keywords: The objective of this paper is to provide a behavioral explanation for risk taking in Islamic and
Islamic banks conventional banks. Within the prospect theory framework; we consider the bias of loss aversion
Risk and mental accounting to explain the risk behavior in Islamic and conventional banks in the
Prospect theory MENA region. We use the Fishburn's (1977) risk measure and the Kendall’s t to test the prospect
Loss aversion
theory predictions. Several measures of performance and risk are used as target level. The results
Mental accounting
for the two types of banks are too similar and provided evidence for Fishburn's (1977) risk
measure and Tversky and Kahneman's (1992) cumulative prospect theory. Banks above the target
level tend to show risk aversion behavior, while the banks listed below tend to be risk-oriented.
This finding provides evidence for the loss aversion bias and mental accounting.

1. Introduction

The Islamic financial system is growing widely over the last thirty years. Islamic finance has become one of the segments of the
financial industry and has occupied a very important place in many countries (Hasan and Dridi 2010). As part of this Islamic sphere,
Islamic banks are becoming increasingly important. They constitute new players whose activity is governed by the principles of the
Islamic Shari'ah, which requires ethical rules in harmony with financial practices and banking products.
While the traditional bank relies on the concept of the interest rate to take advantage of its operations, the Islamic bank has no
access to this mechanism. These Islamic banks have adopted techniques that replace interest income with cash flows from productive
sources (Hassoune, 2009; Tayebi, 2008).
However, Islamic financing, which is motivated by the desire to ensure a justified distribution of wealth and income, does not
appear to be safe from risk. Moreover, this type of banks might to be induced to take on more risk in order to compete with their
conventional counterparts. This subject of risk has always been explained by the traditional finance paradigm, which is based on the
rationality of agents. This rationality is based on perfect information and coherent beliefs. However, classical explanations based on
rationality prove insufficient to explain the risk which is of major importance in the banking activity especially with the recurrence of
the crises of the banking system. The emergence of an innovative research stream called behavioral finance has given rise to a new
explanatory approach based on the cognitive and emotional biases inherent in the behavior of managers within banking institutions.
Indeed, it is only recently that particular attention is given to the study of behavioral biases of managers for the explanation of bank’s
risks. These biases, which are cognitive and emotional, have the power, to influence significantly the decisions of managers.
Behavioral finance models based on cognitive psychology suggest specific features of agent’s behavior, relaxing the rationality
hypothesis (Barberis and Thaler, 2003). In this same context of behavioral finance, another essential assumption is made about the
preferences of investors and how to evaluate risky choices.The prospect theory, introduced by Kahneman and Tversky (1979), is the

⁎
Corresponding author.
E-mail addresses: nousssa_2010@live.fr (I. Ben Salah Mahdi), abbes.mouna@gmail.com (M. Boujelbène Abbes).

http://dx.doi.org/10.1016/j.ribaf.2017.07.111
Received 5 February 2017; Accepted 5 July 2017
Available online 13 July 2017
0275-5319/ © 2017 Elsevier B.V. All rights reserved.
I. Ben Salah Mahdi, M. Boujelbène Abbes Research in International Business and Finance 45 (2018) 577–587

most famous of these theories.This theory is essentially based on the criticism of the expected utility theory as a descriptive model for
decision-making under risk.The prospect theory suggests several important features. First, utility is defined on the gains and losses,
and not in the final wealth (mental accounting). Second, institutions are rather more averse to losses than they are attracted to gains
(lossaversion). Third, institutions are risk seekers in the area of losses, and risk aversion in the area of earnings (asymmetric risk
preference). finally, this theory suggests the feature of non-linear transformation of probabilities, where institutions evaluate extreme
events in the manner that they overestimate low probabilities and underestimate high probabilities (probability weight function)
(Kahneman and Tversky, 1979).
In this paper, we focus on the banking industry, but in a specific framework, conventional and Islamic banks, where risk taking
behavior can become adverse, generating excessive risks for Islamic banks. The specificities of these banks, in the sense that they are
guided by the principles of Shari'a (prohibition of interest, profit and loss sharing, etc.), may foster excessive risk taking, affecting the
perception of risk by the bankers. Therefore, an investigation of this risk perception in a behavioral finance framework is suitable.
In this context, we examine the relevance of the prospect theory as well as the risk measure of Fishburn (1977) to explain risk
taking behavior in the Islamic and conventional banking system. Admitting that risk-taking decisions are influenced by human
subjective judgment as well as perception of risk, then it seems relevant to engage in the behavioral perspective in order to better
study and understand this process. We empirically study risk taking in the banking sector within the framework of the cumulative
prospect theory.
As there are relatively few studies on the risk behavior of Islamic banks, this study contributes to widening the scope of the
literature by providing empirical results on the behavioral biases affecting risk taking of Islamic banks. To our knowledge, this is the
first study that considers the loss aversion bias and mental accounting to explain the risk behavior of Islamic and conventional banks
in a comparative framework.
This paper is organized in the as follow: Section 2 present a literature review, section describes data and methodology, the
empirical results based on the prospect theory are presented in Section 4 and finally Section 5 concludes the paper.

2. Literature review

Fishburn (1977) suggests a portfolio allocation model, alternative to “mean-variance”, where risk is defined as the dispersion of
returns below the target. The model is motivated by the idea that portfolio managers generally associate a risk with the failure to
achieve a certain benchmark level of performance.
In the same way, Kahneman and Tversky (1979) propose the prospect theory as an alternative to the theory of expected utility in
decision making. They show that when individuals are confronted with events involving a set of choices, the real response of these
individuals does not necessarily follow the rational calculation of expected utility on the basis of the objective notion of risk-return
(Kahneman and Tversky, 2000). The main contribution of Kahneman and Tversky, focuses on a set of choices presented to individuals
as well as on the events underlying this set of choices. Their findings largely resultfrom laboratory experiments, in which participants
were generally asked to choose a preferred alternative (with particular risk-return characteristics) from a set of choices.
First, prospect theory was developed to explain individual choices based on predefined losses or gains and fixed probabilities
associated with other alternatives (Kahneman and Tversky, 1979, 2000). As the probability of outcomes for true organizational
choices can’t be clearly determined (March and Shapira, 1987) and external factors are likely to be present in organizations (Bromiley
et al., 2001), the question Therefore, is whether the principles of prospect theory can be applied to organizational parameters?
Bowman (1980) introduce the risk-return paradox in strategic management (in the context of underperforming firms that also
have higher risks). Then, some testable hypothSesis is suggested by Kahneman and Tversky (1979) framework: the decision makers
are risk-seeking in a below target level and risk-averse above given target level. Payne and al. (1980 and 1981) confirm the Kah-
neman and Tversky results and show that basing on prospect theory, an individual can exhibit different level of risk aversion.
The prospect theory’s feature requires that risk attitude is determined by the outcome relation to a reference point of a grouping of
other perspectives, among which a choice must be made. This framework acts as a mental filter to organize preferences for other
modes of action (eg a strategic choice to invest or not to invest).
Barberis (2013) findings are of particular interest showing an important insight into the prospect theory in firms. The first idea
stipulates that choices often occurs in relation to the status quo as a point of reference. The status quo may not only reflect a situation
at some point in time but may also include a trajectory as an income stream or an expected trend of market share in the case of
unchanged behavior. The second idea states that, with the status quo, defined as a reference point, the perception of choices will be
associated with safe gains (following a particular framework) and will generally be favored independently of the expected values.
Safe gains, even if they are low, are systematically preferred to probabilistic results where losses are a possibility of realization (even
with a low probability), reflecting risk aversion (losses are more “threatening” than earnings). In the context of managerial man-
agement, the concept of certainty is different from that which prevails in the context of a laboratory experiment. Here, managers have
a propensity to think that they can at least partially control the future of their business in the face of a difficult strategic event, such as
a threat from the environment (March and Shapira, 1987). More precisely, if the expected result of a particular reaction at the firm
level to a difficult strategic event is perceived as a secure gain (compared to other responses, losses are a possibility), then risk
aversion will be selected. However, if the perceived result of a particular reaction could be the avoidance of safe losses, even if the
likelihood of such a success would be low (compared to alternative responses, losses are a certainty) then the behavior of risk research
prevail.
This proposition was empirically tested by Fiegenbaum (1990) using a sample of 85 US industries composed of about 3300 firms.
The results show that companies below the industry target are looking for risk, while those above are risk averse.

578
I. Ben Salah Mahdi, M. Boujelbène Abbes Research in International Business and Finance 45 (2018) 577–587

Focusing on the banking context, much research is made to analyze the main factors influencing bank risk-taking decisions.
According to Fishburn (1977), Kahneman and Tversky (1979) state that the prospect theory constitutes an important factor in
decision making process in the commercial banking industry.
Johnson (1994) makes his research of risk-taking behavior in American commercial banks and provide that the majority of banks,
who their returns are below the reference point, are generally less risk averse.
Godlewski (2004) conduct an empirical study to analyze risk taking in the commercial banking sector of the South-East Asia and
South and Latin America countries for the period 1996–2001. He concludes that cumulative prospect theory explains risk behaviors
within banks. Its results show risk-averse behavior for banks located above the target. Godlewski (2007) shows that banks located
above the target level (assessed by different measures of return and risk) tend to exhibit unfavorable risk behavior, unlike banks that
are below the target level are not necessarily risk seekers.
On a sample composed of 10 commercial banks that are listed on the Tunisia Stock Exchange over theperiod 1999–2008, Zribi and
Boujelbène (2012) conclude that in below target level, the results confirm the prospect theory since the distance from target is
generally negatively associated with the dispersion fromthe mean. In above target level, they show that distance from targetis
generally positively associated with the dispersion from the mean.
In the context of the Islamic banking system, research on excessive risk-taking and leverage appears to be very limited. The
principal purpose of the study is to examine the power of the prospect theory and Fishburn’s measure of risk to explain variability of
accounting measures and to analyze alternative definitions of target outcomes in order to differentiate the risk-taking behavior in the
Islamic and conventional banking industry.
It is of great importance to integrate Islamic banks into this study, since they present usually risky aspect that originated in the
type of contracts put in practice, obeying Sharia principles, such as the Moudharaba and Moucharaka contracts. However, risk taking
in the Islamic banking system is not well explored in the context of prospect theory. The contribution of this paper is to investigate the
behavioral biases that affect risk-taking of Islamic banks as compared to conventional banks.

3. Data and methodology

3.1. Data and variables definition

In this study, we use data relative the Islamic and conventional banks of the MENA region. These data were obtained from the
Bankscope database for the period 2005–2013. Our sample is composed of 88 conventional banks and 41 Islamic ones as shown in
Table 1.
We use several measures of return and risk as proxies of target level. Concerning the return measures, we follow the existing
literature from commercial banking industry. We use return on equity (ROE) (reflects the shareholder point of view) and return on
assets (ROA) (designates the management point of view) (Blair and Heggestad, 1978; Koehn and Santomero, 1980; Sealey, 1983). The
ratio of Interest income to total loans (IINTL) and the interest income to total operating income ratio (IINTOI) are used in this paper to
give a broader perspective of bank’s credit activities. It should be noted that Islamic banks do not treat interest, which let to consider
for this type of bank fixed profit charges or financing charges as interest income (Alam and Tang (2012)).
Concerning the risk measures, apart from the standard deviation of the return variables, we also investigate accounting measures
of risk: impaired loans to gross loans ratio (IPLGL), loan loss reserves to impaired loans ratio (LLRIPL), loan loss reserves to gross loans
ratio (LLRGL), and net loan to total assets ratio (NLTA). IPLGL and LLRIPL are bank return measures while accounting based measures
LLRGL, NLTA and LLPTL are employed to reflect credit risk (Drake and Hall 2003; Charnes et al., 1990; Hassan, 1993; Mansur et al.,
1993; Mansur and al., 1993). We choose These variables according to Agusman et al., (2008), since they can be useful in the study
and analysis of different banking systems. We employ also the ratio of equity to total assets (EQTA) following Hughes and Mester

Table 1
Geographical repartition by type of banks.

Country Conventional banks Islamic banks Total

Bahrain 9 10 19
Iraq 1 1 2
Algeria 2 1 3
Egypt 9 2 11
Jordan 9 3 12
Kuwait 5 3 5
Qatar 5 2 7
SaudiArabia 9 3 12
The Palestine 1 1 2
Tunisia 4 1 5
Turkey 12 3 15
United ArabEmirates 13 5 18
Yemen 2 2 4
Pakistan 7 4 11
Total 88 41 129

579
I. Ben Salah Mahdi, M. Boujelbène Abbes Research in International Business and Finance 45 (2018) 577–587

Table 2
Descriptive Statistics of Islamic and Conventional Banks variables.

Panel A: Islamic banks

Mean Median SD Min Max

Performance Measures
ROE Net income/Equity 0,10 0,10 0,45 −3,00 7,37
ROA Net income/Total assets 0,01 0,01 0,08 −0,88 0,26
IINTL Interest income/Total loans 0,55 0,05 2,98 −2,23 31,03
IINTOI Interest income/Total operating income 0,59 0,70 0,50 −4,95 2,13
RiskMeasures
IPLGL Impaired loan/Gross loan 1,24 0,07 6,03 0,00 100,83
LLRIPL Lloss reserves/Impaired loan 1,45 0,63 5,78 −0,05 91,5
LLRGL Lloss reserves/Gross loan 0,09 0,04 0,18 0,00 1,01
NLTA Net loan/Total assets 0,71 0,54 1,52 0,00 14,65
EQTA Equity/total assets: capital risk 0,23 0,16 0,21 −0,09 1,19
LLPTL Loanloss provision/Total loans:credit risk 0,12 0,01 0,72 −0,02 95,26
TLDEP Loan/Deposits: liquidity risk 2,47 0,71 10,18 0 163,4

Panel B: Conventional banks

Mean Median SD Min Max

Performance Measures
ROE Net income/Equity. 4,21. 0,14. 42,33. −232,95. 985,02.
ROA Net income/Total assets 0,63 0,02 6,74 −19,29 169,32
IINTL Interest income/Total loans 0,07 0,05 0,05 −0,03 0,63
IINTOI Interest income/Total operating income 0,67 0,69 0,15 −0,41 1,45
RiskMeasures
IPLGL Impaired loan/Gross loan 0,22 0,05 1,23 0,00 18,40
LLRIPL Lloss reserves/Impaired loan 2,05 0,93 8,28 0,03 84,13
LLRGL Lloss reserves/Gross loan 0,16 0,04 0,96 0,00 14,36
NLTA Net loan/Total assets 0,73 0,54 4,19 0,02 117,64
EQTA Equity/total assets: capital risk 0,14 0,12 0,08 0,01 1,00
LLPTL Loanloss provision/Total loans: credit risk 0,01 0,01 0,02 −0,07 0,25
TLDEP Loan/Deposits: liquidity risk 1,01 0,68 6,21 0,02 173,59

(1993) and Mester (1996)) as capital risk variable. Finally, we use the ratio of total loans to deposits (TLDEP) for indicate liquidity
risk. The definition of variables and their descriptive statistics are given in Table 2.

3.2. Methodology

To test the contribution of the prospect theory to explain the risk taking in Islamic and conventional banks, we use the Fishburn
(1977) measure to estimate the risk and the Kendall t to account for the phenomenon of framing.
Fishburn (1977) suggests that risk is not necessarily a measure of dispersion around an expected value, but rather a function of
distance from a reference point that can be represented by the following equation:
t
R(t ) = ∫ (t − x )α dF (x )
−∞ (1)
Where R(t) is the measure of risk. t is the target level or reference point. The medians for different performance and risk measures
represent target levels or benchmarks for the bank.α is the sensitivity to the deviation of the target, α > 0. F (x) is the density
function of the probability of x.
The positive parameter α is a measure of the attitude of an individual towards the results below the target. The Fishburns’
measure of risk predicts that the smaller these values (the greater the distance from target), the more appealing larger standard
deviations should be. Our analysis depends on time average and their standard deviations measures, as well as median of risk and
return variables. The medians of the employed measures represent the target levels or the reference points for the bank.
We also study the framing phenomenon by using the t measure of Kendall, which tests the correlations between the risk and
return measures in the areas of gain compared to loss and vice versa. We note that the Kendall’s t is used to measure the correlations
between these variables within Islamic and conventional banks. The possible values of Kendall’s t vary between +1 (perfect positive
correlation) and −1 (perfect negative correlation). This can be argued by the fact that the distance from the target will be negative
for all banks that fall below this reference point. In addition, the greatest distances of the target are associated with more negative
values of DTROA (The distance from the ROA target), DTROE (The distance from the ROE target), DTIINTL (The distance from the
IINTL target) and DTIINTOI (The distance from the IINTOI target) (Smaller values). The prospect theory suggests that banks that are
below the target take excessive risks. The Fishburn risk measure admits that a smaller value, (greater distance from the target), its

580
I. Ben Salah Mahdi, M. Boujelbène Abbes Research in International Business and Finance 45 (2018) 577–587

standard deviation is higher, and therefore will be more attractive. Thus, when Kendall's t is negative, this denotes a situation below
the target. The null and alternative hypotheses are as follows:

H0: tBT < 0

HA: tBT ≥ 0

Where tBT is the Kendall’s t correlation between the distance of the target and the standard deviation for the banks below the target.
The rejection of the null hypothesis suggests that the distance below the target (related to the Fishburn risk measure) is associated
with greater variability in the observed variable.
Above the target, the prospect theory suggests risk aversion. Indeed, a greater distance above the target should induce less risk-
taking (eg, less variability in return) and therefore a positive correlation.
However, the Fishburn risk measure is a measure that is strictly below the target. Thus, above the target, the Fishburn measure of
the risk takes the value of zero. In addition, distance from the target should have little impact on variability.
The null and alternative hypotheses are:

H0: tAT = 0

HA: tAT ≠ 0

Where tAT shows the correlation of Kendall's t between the distance of the target and the standard deviation for the banks above the
target.
The rejection of the null hypothesis suggests that the distance greater than the target (whose Fishburn measure of the risk is zero)
is related to the degree of variability of the observed variable. If the null hypothesis in the first equation is rejected while the null
hypothesis in the second equation is accepted, it would mean that the distance of the target has a potential explanatory power below
the target but not above it. These results support the Fishburn risk measure and therefore the prospect theory
Kendall's t interpretation in accordance with the prospect theory is presented in Fig. 1.

4. Empirical results

According to the cumulative prospect theory, the agent does not reason with respect to the final wealth but with respect to the
change in wealth. Thus, we study the risk-taking behavior on the one hand in the area of gain (return variables) and on the other hand
in the area of loss (risk variables).
For each zone, we split the sample into two areas “above and below” corresponding respectively to banks above and below the
target level.

Fig. 1. Prospect Theory: gain vs loss domain.

581
I. Ben Salah Mahdi, M. Boujelbène Abbes Research in International Business and Finance 45 (2018) 577–587

Table 3
Correlations between the standard deviation and the distance from the reference point (gain domain) for Islamic banks.

ROA (Zone 1) ROE (Zone 2) IINTL (Zone 3) IINTOI (Zone 4)

Above −0.041* −0.055* −0.046* −0.032*

(0.068) (0.088) (0.082) (0.054)
Below 0.0183* 0.018* 0.0207* 0.035*
(0.085) (0.070) (0.056) (0.056)

Note: ***, **, * respective significances at the 1%, 5%, 10%.

4.1. Risk taking behavior in the gain’s zone

In Tables 3 and 4 we calculate the Kendall’s t correlation coefficients for the Islamic and conventional banks between the standard
deviation of the variable and the distance from the target level corresponding to each gain zone.
Regarding the correlation results for zones 1–4, which relates to the gain domains for Islamic banks, we observe that the Kendall’s
coefficient is always significant. For areas “above”, the coefficients are negative.
This result indicates that Islamic banks above the target levels in terms of ROA, ROE, IINTOI and IINTL exhibit risk aversion
behavior, since the standard deviation and distance for the median are negatively correlated. This may reflect a more defensive
attitude on the part of successful Islamic banks. Indeed, being above the target, these banks want to preserve the best position, and
therefore exhibiting risk aversion.
For banks below the target, all findings are also significant and show positive signs. This reflects excessive risk-taking behavior by
banks below the target. For the zone (ROE) concerning shareholders, we can conclude that this excessive risk taking may be due to
additional pressure on the management of these Islamic banks to please their shareholders. The same reasoning is still valid for the
manager’s side, since the return on assets (ROA), also presents a positive sign for banks located below the median. Therefore, the
prospect theory and the Fishburn measure of risk are confirmed for all zones. Indeed, we have shown risk aversion behavior for
Islamic banks above the target and a risk-seeking behavior for Islamic banks that are below the median.
The results found in this table shows that conventional banks, above the target, exhibit risk aversion behavior for all subareas,
except for the third. Indeed, the distance between the standard deviation of the variables (ROA), (ROE) and (IINTOI) and their
medians are negatively correlated. For conventional banks below the target level, the relationship between these two measures is not
significant.

4.2. Risk taking behavior in loss’s zone

Table 5 presents the Kendall’s t in the risk domain for Islamic banks.
Concerning the Islamic sample, banks above the target show a significant and negative Kendall’s t for all zones, except those of
(LLRGL) and (EQTA). In terms of potential losses, presented by IPLGL, bankers exhibit risk aversion behavior. Indeed, a level of IPLGL
above the target induces a higher attitude of risk aversion, since potential losses can lead the bank to default. However, the Kendall
correlation’s coefficients for zones below are all positive and significant, reflecting excessive risk-taking by underperforming banks.
Table 6 presents the Kendall’s t in the risk domain for conventional banks.
The results found for conventional banks are almost identical to those found for Islamic banks in the risk zone. We conclude that
most powerful conventional banks, who are located above the target, exhibit risk aversion behavior. However, the Kendall correlation
coefficients for the zones below are no significant.

4.3. Risk taking behavior in the domain of gains vs losses

In Tables 7 and 8, we propose to cross domains (gains vs losses) to study the predictions of prospect theory in Islamic and
conventional banking respectively, in an attempt to investigate the framing issue which is one of the crucial features of prospect
theory. Indeed, the negatively supervised problems reduce the risk aversion and encourage the search for risk. This behavior is mainly
attributed to loss aversion bias. Since losses are more threatening than gains, it appears that managers can follow conservative
strategies when they are in the context of positive changes in wealth relative to the benchmark, and risky strategies when they face to

Table 4
Correlations between the standard deviation and the distance from the reference point (gain domain) for conventional banks.

ROA (Zone 1) ROE (Zone 2) IINTL (Zone 3) IINTOI (Zone 4)

Above −0.025*** −0.009*** 0.004 −0.034***

(0.003) (0.005) (0.925) (0.001)
Below −0.0219 −0.0082 0.025 0.0210
(0.524) (0.819) (0.408) (0.551)

Note: ***, **, * respective significances at the 1%, 5%, 10%.

582
I. Ben Salah Mahdi, M. Boujelbène Abbes Research in International Business and Finance 45 (2018) 577–587

Table 5
Correlation between the standard deviation and the distance from the target (loss domain) for Islamic banks.

IPLGL Zone 5 LLRIPL Zone 6 LLRGL Zone 7 NLTA Zone 8 EQTA Zone 9 LLPTL Zone 10 TLDEP Zone 11

Above −0.144* 0.013* 0.010* −0.016* 0.030* −0.042* −0.093*

(0.064) (0.081) (0.092) (0.072) (0.062) (0.068) (0.069)
Below 0.017* 0.017* 0.011* 0.018* 0.018* 0.006* 0.043
(0.064) (0.070) (0.077) (0.073) (0.068) (0.086) (0.248)

Note: ***, **, * respective significances at the 1%, 5%, 10%.

Table 6
Correlations between the standard deviation and the distance from the target (loss domain) for conventional banks.

IPLGL (Zone 5) LLRIPL (Zone 6) LLRGL (Zone 7) NLTA (Zone 8) EQTA (Zone 9) LLPTL (Zone 10) TLDEP (Zone 11)

Above −0.078 −0.014* −0.007* −0.008* −0.014* −0.029** −0.005*

(0.363) (0.093) (0.094) (0.079) (0.071) (0.050) (0.087)
Below 0.008 0.021 0.019 −0.014 0.024 0.029 −0.118
(0.756) (0.384) (0.446) (0.704) (0.434) (0.349) (0.970)

Note: ***, **, * respective significances at the 1%, 5%, 10%.

negative changes in wealth relative to the same point.

We test this behavior in banks of our sample as follows: we calculate the Kendall correlation’s coefficient between the standard
deviations of the return’s measures (SDROE, SDROA, SDIINTL and SDIINTOI) and the distance to the median of the loss’s measures
(IPLGL, LLRIPL, LLRGL, NLTA, EQTA, LLPTL et TLDEP) corresponding to the zones (5–11). The analysis of the results presented in
Table 7 shows significant Kendall’s t only for the area below for zones 5, 6, 7, 9 and 10. For zones 8 and 11, we observe significance
only for the area above the target suggesting a risk-taking behavior by Islamic banks underperforming in terms of potential losses
IPLGL and their coverage LLRIPL and LLRGL as well as in terms of EQTA and LLPTL. Indeed, being below the reference, leads to an
aggressive risk taking within the bank. This reflects an offensive strategy by underperforming Islamic banks. As for the results of
banks above the benchmark, we observe significantly negative Kendall’s coefficients between the distance to the target in terms of
NLTA and TLDEP and the standard deviations of the return measures.
This reflects that the banks that have high ratios of loans in relation to total assets or the total deposits tend to take less risk as
measured by the standard deviation of return measures. This risk aversion behavior is explained by the fact that even if this volume of
loans represents potential revenues for banks, it could also turn into impaired loans which increasing the bank's risk of default. So, the
signs of Kendall's t on the significant correlations confirm the predictions of the prospect theory thus translating a risk taking by the
leaders of the Islamic banks that are located above the target. We note on the other hand a risk aversion on the side of the leaders of
the performing banks.
In Table 8, we calculate the Kendall correlation’s coefficient between the standard deviations of return measurements (SD ROE, SD
ROA, SD IINTL et SD IINTOI) and the distance of losses measured by their medians corresponding to the zones (5–11).
By analyzing the correlation coefficients for the banks below the target, we observe that in terms of potential losses IPLGL and
their coverages LLRIPL and LLRGL, the Kendall’s t shows positive signs indicating a risk taking behavior. However, for banks above
the benchmark, we observe significantly negative coefficients of the Kendall coefficient between the target distance in terms of NLTA,
LLPTL and TLDEP and the standard deviations of the return measures. This implies that banks with high ratios of loans in relation to
total assets NLTA or relative to total deposits TLDEP and those that have a high level of loan loss provisions LLPTL tend to take less
risk measured by the standard deviation of the performance variables. This risk aversion is explained by the fact that even if this
volume of loans represents potential revenues for banks, it could also turn into impaired loans, thus increasing the bank's risk of
default. So, the prospect theory perfectly explains the bank risk-taking leaders of conventional banks.

4.4. Risk taking behavior in the domain of loss vs gain

In Tables 9 and 10, we propose to cross domains of loss vs. gain to study the predictions of prospect theory in Islamic and
conventional banking respectively. Indeed, negatively controlled problems reduce risk aversion because of the loss aversion bias. In
this step, we reverse the measurements, presenting the Kendall’s coefficients between the standard deviations of the loss measure-
ments and the distance to the medians of the measurements of gains (zones 1–4). Looking at the results in Table 9, for Islamic banks
above the target, we note that the correlation coefficients corresponding to return on equity (ROE) and return on assets (ROA) are all
negative. This implies that performing Islamic banks exhibit risk-averse behavior on the side of shareholders and managers, as the
relationship between distance to target and standard deviation of loss measures is negative. This can also be interpreted as un-
favorable risk behavior, implying more prudent behavior. The Islamic banks above the benchmark adopt a conservative strategy since
they show a defensive behavior. This result confirms the predictions of the prospect theory.
As regards zone 3 relating to IINTL, we observe a different behavior, because the Kendall correlation is positive for the banks
situated above the target. This indicates excessive risk-taking behavior in Islamic banks. For zone 4 (IINTOI), the relationship is not

583
 I. Ben Salah Mahdi, M. Boujelbène Abbes

Table 7
Correlations between standard deviation and distance from the target (gain vs loss domain) for Islamic banks.

IPLGL (Zone 5) LLRIPL (Zone 6) LLRGL (Zone 7) NLTA (Zone 8) EQTA (Zone 9) LLPTL(Zone 10) TLDEP(Zone 11)

Below Above Below Above Below Above Below Above Below Above Below Above Below Above

SDROE −0,02 0,37*** 0,18 0,04** 0,01 0,30*** −0,18** −0,14 0,14 0,00 (0.826) 0,00 0,25*** −0,09* (0.056) −0,11
(0.811) (0.00) (0.317) (0.061) (0.945) (0.00) (0.019) (0.549) (0.160) (0.979) (0.000) (0.125)

584
SDROA −0,01 0,14** −0,09 0,02*** 0,10 0,32*** −0,61*** 0,01 0,67 0,33*** 0,06 0,26*** −0,25*** −0,11
(0.907) (0.042) (0.270) (0.000) (0.236) (0.000) (0.000) (0.874) (0.980) (0.000) (0.436) (0.000) (0.001) (0.1924)
SD IINTL 0,38 0,03* 0,53 0,04* 0,99 0,09* −0,95*** −0,23 −0,81 0,19*** 0,38 0,09* −0,28* (0.059) −0,07
(0.124) (0.083) (0.622) (0.079) (0.950) (0.090) (0.000) (0.989) (0.630) (0.000) (0.118) (0.092) (0.217)
SDIINTOI 0,03 0,07* 0,01 0,06* 0,56 −0,01 −0,24*** −0,42 0,09 0,71*** 0,35 0,01 (0.883) −0,03* (0.065) 0,33 (0.603)
(0.653) (0.064) (0.833) (0.052) (0.985) (0.881) (0.000) (0.632) (0.150) (0.000) (0.230)
Research in International Business and Finance 45 (2018) 577–587
 I. Ben Salah Mahdi, M. Boujelbène Abbes

Table 8
Correlations between standard deviation and distance from the target (gain vs loss domain) of conventional banks.

IPLGL (Zone 5) LLRIPL (Zone 6) LLRGL (Zone 7) NLTA (Zone 8) EQTA (Zone 9) LLPTL (Zone 10) TLDEP (Zone 11)

Above Below Above Below Above Below Above Below Above Below Above Below Above Below

SDROE 0,08 0,03* −0,08 0,04* 0,42 0,03* −0,03* (0.057) 0,18 −0,25 −0,13 −0,14***(0.004) 0,173 −0,02* (0.074) 0,09
(0.696) (0.054) (0.289) (0.080) (0.320) (0.056) (0.540) (0.300) (0.114) (0.01) (0.108)

585
SDROA −0,03 0,13** −0,08 0,12** 0,02 0,11** −0,15*** 0,22 0,56 0,15 −0,06* (0.087) 0,10 −0,07* (0.067) 0,17
(0.507) (0.013) (0.111) (0.021) (0.700) (0.027) (0.002) (0.850) (0.320) (0.604) (0.655) (0.970)
SD IINTL 0,29 0,05* −0,13 0,08* 0,37 0,05* −0,43*** −0,42 −0,07 −0,13 −0,10* (0.081) 0,07 −0,18*** −0,41
(0.306) (0.054) (0.526) (0.088) (0.985) (0.063) (0.000) (0.523) (0.232) (0.404) (0.130) (0.001) (0.250)
SDIINTOI 0,23 0,09** −0,01 0,10* 0,45 0,08* −0,06* (0.084) −0,05 0,26 0,08 −0,03* (0.054) 0,19 −0,01* −0,06
(0.360) (0.100) (0.765) (0.061) (0.258) (0.069) (0.317) (0.298) (0.115) (0.650) (0.0897) (0.251)
Research in International Business and Finance 45 (2018) 577–587
I. Ben Salah Mahdi, M. Boujelbène Abbes Research in International Business and Finance 45 (2018) 577–587

Table 9
Correlations between standard deviation and distance from the target (domain of loss vs gain) for Islamic banks.

ROE (Zone 1) ROA (Zone 2) IINTL (Zone 3) IINTOI (Zone 4)

Above Below Above Below Above Below Above Below

SD IPLGL −0.0451 (0.768) −0.1488 −0.1343* (0.079) −0.0363 0.5751*** 0.2157 −0.6996 −0.0156
(0.707) (0.515) (0.000) (0.560) (0.850) (0.779)
SD LLRIPL −0.0511* −0.0021 −0.0412 (0.0624) 0.071 (0.285) 0.0193* (0.081) −0.0494 0.0563 −0.0161
(0.0543) (0.974) (0.461) (0.502) (0.810)
SD LLRGL −0.2809** −0.0129 −0.1470 (0.288) −0.2015 0.2200 (0.109) 0.0302 0.1478 0.1349 (0.316)
(0.039) (0.820) (0.560) (0.592) (0.286)
SD NLTA −0.3840*** −0.0520 −0.3674*** −0.1591 0.1700** −0.2032 0.1097 0.2592 (0.301)
(0.000) (0.523) (0.000) (0.349) (0.012) (0.211) (0.108)
SD EQTA −0.1039* (0.067) 0.0393 −0.0690* (0.072) −0.2751 0.2190** 0.1200 0.0904 0.0934 (0.146)
(0.542) (0.546) (0.013) (0.161) (0.314)
SD LLPTL −0.0679 (0.657) 0.2103 −0.0176* (0.090) −0.2330 0.3273** 0.0363 0.1309 0.1762 (0.401)
(0.369) (0.157) (0.028) (0.515) (0.391)
SD TLDEP −0.1100 (0.471) −0.0285 −0.0755* 0.0143 0.5819*** 0.2224 0.1039 −0.2288
(0.608) (0.0622) (0.797) (0.000) (0.450) (0.496) (0.236)

Table 10
Correlations between standard deviation and distance from the target (loss vs gain domain) for conventional banks.

ROE (Zone 1) ROA (Zone 2) IINTL (Zone 3) IINTOI (Zone 4)

Above Below Above Below Above Below Above Below

SD IPLGL −0.027*(0.083) −0.129 −0.014* (0.0911) −0.133 −0.173* 0.089 −0.037* (0.077) 0.013
(0.900) (0.560) (0.074) (0.416) (0.730)
SD LLRIPL −0.601***(0.000) −0.044 −0.7897*** (0.000) −0.121 −0.328* −0.103 −0.712*** (0.000) 0.015
(0.226) (0.650) (0.094) (0.504) (0.690)
SD LLRGL −0.078* (0.054) −0.092 −0.116* (0.064) −0.120 −0.217* 0.172 −0.215* (0.091) −0.053
(0.314) (0.200) (0.086) (0.560) (0.155)
SD NLTA −0.038*(0.065) −0.015 −0.049* (0.080) 0.224 −0.417*** −0.080 −0.184*** (0.000) −0.012
(0.782) (0.563) (0.000) (0.147) (0.828)
SD EQTA −0.093 (0.083) −0.120 −0.257***(0.000) −0.180 −0.044* −0.033 −0.055*(0.075) −0.068
(0.549) (0.300) (0.099) (0.476) (0.144)
SD LLPTL −0.041*(0.073) −0.252 −0.083* (0.077) −0.283 −0.658*** −0.077 −0.005* (0.0934) −0.234
(0.650) (0.230) (0.000) (0.196) (0.652)
SD TLDEP −0.020* (0.068) 0.013 −0.376***(0.000) 0.096 −0.182*** −0.020 −0.156*** (0.001) 0.077
(0.804) (0.262) (0.000) (0.692) (0.133)

significant. However, the Kendall correlation’s coefficients for areas below the target are all non-significant.
On the whole, we observe that within a framework of loss, the behavior of Islamic banks below a target level is affected by a risk-
taking attitude. On the contrary, in a gain setting, the behavior of banks above a target level is influenced by an attitude of risk
aversion.
In Table 10 we calculate the Kendall correlation coefficient between the standard deviations of loss measurements and the
distance to the return measurements for the conventional banks (zone 1–4).
By observing the results in Table 10, we notice Kendall’s t negative and significant only for the banks situated above the target for
zones 1, 2, 3 and 4. For zone 1, relative to return on equity ROE, it indicates a negative relation between the distance to the ROE
target and the standard deviation of the measures of risk. We conclude for risk aversion behavior on the part of the shareholders of
performing conventional banks. For zone 2, which corresponds to the ROA, the results indicate risk aversion behavior rather on the
direction side, since the relationship between the distance to the ROA target and the standard deviation of the measurements of losses
is negative.
This risk aversion behavior is also recorded for zones 3 and 4, corresponding respectively to the IINTL and IINTOI zones. Indeed,
these two variables represent two proxies for the credit activities, which is the first source of banking gain. Thus, any expected loss,
materialized by the negative relationship between the standard deviations of losses measurements and the distance to the median,
seems to discipline the risk-taking behavior in performing banks. This behavior is consistent with the prospect theory which suggests
risk aversion for banks above the target level.

5. Conclusion

In this paper, we have examined the risk-taking behavior of Islamic and conventional banks under the behavioral finance.
Particularly, we consider the prospect theory to explain the CEO risk taking behavior. Indeed, several studies show that absolute

586
I. Ben Salah Mahdi, M. Boujelbène Abbes Research in International Business and Finance 45 (2018) 577–587

estimates based on final wealth are limited and the perception of gain or loss has become a requirement for understanding the risk-
taking process in banks. The cumulative prospect theory provides an alternative framework in the analysis of risk-taking, particularly
for excessive risk-taking in banks that remains the main cause of their failure. Although the literature dealing with these issues of risk-
taking behavior in banks is still extremely rare, the results of this study provide an empirical insight into the usefulness of the
behavioral framework for the analysis of risk-taking in Islamic and conventional banks.
We use several measures of return and risk as proxies of target level. Concerning the return measures, we use (ROE), (ROA),
(IINTL) and (IINTOI). Concerning the risk measures, apart from the standard deviation of the return variables, we also investigate
accounting measures of risk (IPLGL), (LLRIPL), (LLRGL), (NLTA), (LLPTL), (EQTA) and (TLDEP).
For a sample of Islamic and conventional banks, our results provide evidence for Fishburn's (1977) risk measure and Tversky and
Kahneman's (1992) cumulative prospect theory. Indeed, we note that banks above the target level tend to show risk aversion be-
havior, while the banks located below tend to be risk-oriented. In other words, we conclude that within a loss framework, being
below a target level seems to influence the risk of the bank by encouraging a risk-seeking. in a gain setting, being above a target level
has a significant impact on the risk aversion of bankers.

References

Agusman, A., Monroe, G.S., Gasbarro, D., Zumwalt, J.K., 2008. Accounting and capital market measures of risk: evidence from Asian banks during 1998–2003. J. Bank.
Finance 32 (4), 480–488.
Alam, N., Tang, K.B., 2012. Risk-Taking behavior of islamic banks: application of prospect theory. Qualitative Res. Financial Markets 4 (2/3), 156–164.
Barberis, N., Thaler, R., 2003. A Survey of Behavioral Finance Handbook of The. In: In: Constantinides, G.M., Harris et, M., Stultz, R. (Eds.), Economics of Finance, vol.
1 B Elsevier (chap 18).
Barberis, N., 2013. Thirty years of prospect theory in economics: a review and assessment. J. Econ. Perspect. 27, 173–196.
Blair, R.D., Heggestad, A.A., 1978. Bank portfolio regulation and the probability of bank failure. J. Money Credit Bank. 10 (1), 88–93.
Bowman, E.H., 1980. A Risk/Return paradox for strategic management, sloan management. Review 21 (3), 17–31.
Bromiley, P., Miller, K.D., Rau, D., 2001. Risk in strategic management research. In: Hitt, M.A., Freeman, R.E., Harrison, J.S. (Eds.), The Blackwell Handbook of
Strategic Management. Blackwell Oxford, pp. 259–288.
Charnes, A., Cooper, W., Huang, Z.M., Sun, D.B., 1990. Polyhedral cone-Ratio DEA models with an illustrative application to large commercial banks. J. Econ. 46,
73–91.
Drake, L., Hall, M., 2003. Efficiency in japanese banking: an empirical analysis. J. Bank. Finance 27, 891–917.
Fiegenbaum, A., 1990. Prospect theory and the risk-return association: an empirical examination in 85 industries. J. Econ. Behav. Org. 14 (2), 187–203.
Fishburn, P.C., 1977. Mean-risk analysis with risk associated with below-target returns». Am. Econ. Rev. 67 (2), 116–126.
Godlewski, C.J., 2004. Bank Risk-Taking in a Prospect Theory Framework: Empirical Investigation in the Emerging Markets ‘Case. Available at SSRN http://www.ssrn.
com SSRN ID588164code347791.
Godlewski, C.J., 2007. An empirical investigation of bank risk-taking in emerging markets within a prospect theory framework: a note. Bank. Bank. Syst. 2 (2), 35–43.
Hasan, M., Dridi, J., 2010. Mise à l’épreuve: Les Banques Islamiques Ont Mieux Résisté A La Crise Que Les Banques Classiques. Finances & Développementpp. 46–47.
Hassan, M.K., 1993. The off-balance sheet banking risk of large US commercial banks. Quarterly Rev. Econ. Finance 33 (1), 51–69.
Hassoune, A., 2009. Finance islamique: un business en plein essor. APECIS.
Hughes, J.P., Mester, L.J., 1993. A quality and risk-Adjusted cost function for banks: evidence on the ‘Too-Big-To-Fail' doctrine. J. Productivity Anal. 4, 293–315.
Johnson, H.J., 1994. Prospect theory in the commercial banking industry. J. Financial Strategic Decisions 7 (1), 73–89.
Kahneman, D., Tversky, A., 1979. Prospect theory: an analysis of decision under risk. Econometrica 47 (2), 263–292.
Kahneman, D., Tversky, A., 2000. Choices, Values, And Frames. In: Kahneman, D., Tversky, A. (Eds.), Cambridge University Press Cambridge, pp. 1–16.
Koehn, M., Santomero, A.M., 1980. Regulation of bank capital and portfolio risk. J. Finance 35 (5), 1235–1244.
Mansur, I., Zangeneh, H., Zitz, M.S., 1993. The association between banks’ performance ratios and market-determined measures of risk. Appl. Econ. 25 (12),
1503–1510.
March, J.G., Shapira, Z., 1987. Managerial perspectives on risk and risk taking. Manage. Sci. 33 (11), 1404–1418.
Mester, J., 1996. A study of bank efficiency taking into account risk-preferences. J. Bank. Finance 20, 1025–1045.
Sealey Jr., C.W., 1983. Valuation, capital structure, and shareholder unanimity for depository financial intermediaries. J. Finance 38 (3), 857–871.
Tayebi, A., 2008. Islamic Finance an Ethical Alternative to Conventional Finance? Association of Charted Certified Accountants.
Tversky, A., Kahneman, D., 1992. Advances in prospect theory: cumulative representation of uncertainty. J. Risk Uncertain. 5 (4), 297–324.
Zribi, N., Boujelbène, Y., 2012. Prospect theory and tunisian banks' risk-Taking, the IUP. J. Behav. Finance IX (1).

587