Using Importance-Performance Matrix Analysis to Evaluate the Financial Performance of American Banks During the Financial Crisis:

Mohamed M. Khalifa Tailab

doi:10.1177/2158244020902079

Using Importance-Performance Matrix Analysis to Evaluate the Financial Performance of American Banks During the Financial Crisis:

Tailab, Mohamed M. Khalifa 2020-01-26 00:00:00 This research applies a technique that identifies areas of improvement that can be addressed by managerial decisions or policy activities. It extends the application of partial least squares structural equation modeling (PLS-SEM) using an importance- performance map analysis (IPMA). The IPMA determines priority factors that should receive management’s attention. The PLS path model was tested by comparing 140 failed U.S. banks with the same number of nonfailed banks from 2006 to 2008. This model assembles 15 indicators with four predecessor constructs (i.e., profitability of 2006, profitability of 2007, risk of 2006, and risk of 2007) and one final target construct (i.e., profitability of 2008). Profitability and risk of 2007 mediate the path of profitability and risk of 2006 and profitability of 2008. The IPMA indicated that failed banks were predisposed to decreasing financial performance in 2008 because of their poor performance in 2006 and 2007. Conversely, nonfailed banks were more likely to experience increasing financial performance in 2008 because of their positive performance in 2006 and 2007. This study indicates that managers who use IPMA to prioritize their financial decisions will obtain useful conceptual insights and are unlikely to be misled. Although IPMA can be conducted on the indicator level as well, this article limits its analysis by focusing on the construct level only. The use of IPMA is ubiquitous in end-user surveys, but its application to banking is still in its embryonic state. For originality, this work prioritizes the application of IPMA using secondary data collected from financial statements to assess the performance of American banks during the crisis. Keywords partial least squares (PLS), structural equation modeling (SEM), importance-performance map analysis (IPMA), banking crisis, measurement invariance, multiple-group analysis performance of banks is very important because they were at Introduction the center of that crisis (M. E. Barth & Landsman, 2010). In Researchers and professionals widely accept that the bank- the wake of the financial crisis, stakeholders are becoming ing sector is the most important component of any financial increasingly concerned with their firm’s financial perfor- system (Georgantopoulos & Tsamis, 2013). Ergo, the stabil- mance; bankers recognize the need to formulate better strate- ity of the banking system contributes to the stability of eco- gies to drive performance but may struggle to determine nomic growth. Thus, the financial crisis in 2007–2008 caused priorities. the biggest economic disruption since the Great Depression Although an unprecedented number of banks collapsed or (Adebambo et al., 2015), shaking the fiscal world with a were bailed out by governments during the crisis (Erkens wave of massive losses (Zaghdoudi, 2013, p. 537). et al., 2012), not all banks across the world performed equally According to the Federal Deposit Insurance Corporation poorly; some banks performed better during the crisis (FDIC), 140 U.S. banks failed in 2009, including several high-profile institutions such as Bear Stearns, Citigroup, Lehman Brothers, Merrill Lynch, and Wachovia. This wide- Lincoln University, Oakland, CA, USA spread failure indicated the weaknesses of the banking system Corresponding Author: (Ayadurai & Eskandari, 2018). Because we now live in a very Mohamed M. Khalifa Tailab, Department of Finance and Investments, interconnected economy, this failure could have been trans- Lincoln University, 401 15th Street, Oakland, CA 94612, USA. mitted to other sectors. Therefore, evaluating the financial Email: mtailab@lincolnuca.edu Creative Commons CC BY: This article is distributed under the terms of the Creative Commons Attribution 4.0 License (https://creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/open-access-at-sage). 2 SAGE Open (Beltratti & Stulz, 2012). To explore this phenomenon, sev- predictors of business failure (Maricica & Georgeta, 2012), 15 eral papers such as Beltratti and Stulz (2012), Serrano-Cinca ratios for each bank were selected to build the initial model. et al. (2014), Adebambo et al. (2015), Cox et al. (2017), and According to the rule of thumb of model evaluation using PLS- Avkiran et al. (2018) investigated the impact of financial cri- SEM (Hair et al., 2016), only six ratios were employed each sis on banks’ performance from a variety of perspectives. year. The model assembles 15 indicators with four predecessor Thus far, however, insufficient attention has been paid by constructs (i.e., profitability of 2006, profitability of 2007, risk these studies to the role of prioritization of managerial deci- of 2006, and risk of 2007) and one final target construct (i.e., sions and policy activities in precipitating the financial crisis. profitability of 2008). Profitability and risk of 2007 mediate the Prioritization at a strategic and operational level is usually path of profitability and risk of 2006 and profitability of 2008. the difference between success and failure. This study helps The importance and performance values of profitability of address this gap in the literature by applying a technique to 2008’s predecessor constructs (i.e., profitability of 2006, profit- help bankers determine priority factors that should receive ability of 2007, risk of 2006, and risk of 2007) create the impor- their attention to prevent a repeat of the failure. tance-performance map of profitability of 2008. In light of this consideration, importance-performance The results plotted in the IPMA indicated that failed banks analysis, which is often called importance-performance map were predisposed to decreasing financial performance in analysis (IPMA) (Ringle & Sarstedt, 2016), has been found 2008 because of their poor performance in 2006 and 2007. to be a useful technique to identify the areas of improvement On the contrary, the profitability of 2006 and 2007 for non- that should be addressed by management activities (Martilla failed banks positioned them for increasing financial perfor- & James, 1977). This approach allows managers to improve mance in 2008. Although the IPMA can be conducted on the their management strategies because it indicates the main indicator level, this article limits its analysis by focusing on factors that require an immediate response (improvement) the construct level only. Finally, this research, to date, is the (Wyród-Wróbel & Biesok, 2017, p. 123). first to apply the IMPA using the secondary data collected The decision to apply this technique was driven by three from financial statements. motivations in particular: (a) IPMA facilitates more rigorous The rest of this article begins by presenting previous management decision-making; (b) IPMA is a powerful tool related work. This is followed by the research method where that can assist managers to set better priorities and better the procedures for predictive model assessment are outlined allocate scarce resources; and (c) having guidelines for per- in clear steps. Findings and discussion follow along with a formance assessment is as valuable to a firm as it is to the brief conclusion. individuals who invest in it (particularly during and follow- ing a financial crisis; Streukens et al., 2018). Literature Review To be clear, the purpose of this study is not to analyze the effect of financial crises on banks’ performance. Rather, it The use of PLS-SEM has received significant attention in introduces the IPMA and how to use it to provide a better several business disciplines (J. R. Barth et al., 2018). This understanding of where management should focus their technique is ubiquitous in marketing and management infor- attention. Reviewing business literature indicates that a mation systems, but it is still in its embryonic state in bank- plethora of papers have used IPMA in banking, such as ing literature because the advantages of this approach have Joseph et al. (2005), Ramayah et al. (2014), and Samar et al. yet to be discovered by the banking discipline (Avkiran, (2017). These papers utilized an end-user satisfaction survey 2018, p. 1). Only a few papers have used PLS-SEM in bank- to measure customers’ perceptions. This article is the first to ing literature. apply IPMA using secondary data collected from financial To my best knowledge, the first application of partial least statements to assess the performance of banks. IPMA was squares discriminant analysis (PLS-DA) for the prediction of applied to a unique data set of 140 failed U.S. banks that the 2008 U.S. banking crisis was done by Serrano-Cinca and closed in 2009 compared with the same number of nonfailed Gutiérrez-Nieto (2013). They compared PLS-DA with eight banks from 2006 to 2008. The U.S. banking data were used algorithms commonly used in bankruptcy prediction. It was as a case study because the crisis started in the United States, indicated that PLS-DA results resemble linear discriminant where many large banks lost most of their equity (Beltratti & analysis and support vector machine results. A particular Stulz, 2012). Thus, as U.S. banks play a major role in today’s advantage of this technique (PLS-DA) is that it is not affected global economy, their performance contributed more heavily by multicollinearity because it has been designed to deal to the crisis than other financial firms around the world. with this issue. They concluded that the interpretability of the It is well known that IPMA is the extension of partial least PLS-DA model was satisfactory. squares structural equation modeling (PLS-SEM; Ringle & The study by Serrano-Cinca et al. (2014) applied a path Sarstedt, 2016). The first requirement of the empirical work is model based on structural equations and logistic regression to to develop the PLS path model. The factors chosen for this investigate the financial symptoms that precede bankruptcy. model are those most relevant to success and failure: profit and They used low profitability, insufficient revenue, or low sol- risk. Because financial ratios are generally utilized as good vency ratios as proxies for symptoms, whereas loan growth Tailab 3 (some of them risky), specialization (real estate concentra- high risk might cause failure, whereas earning consecutive tion), and the pursuit of a turnover-driven strategy neglecting low profit puts management under pressure from sharehold- margin were used as proxies for the causes of these symp- ers. Thus, giving loans to customers is usually associated toms. They found that 5 years before the crisis, distressed with riskier loan practices, which can be measured by risk banks, compared with solvent banks, had the following: ratios. If the customers fail to repay those loans, profitability higher loan growth, higher concentration on real estate loans, will be affected. The initial model of this article is based on higher risk ratios, and higher turnover, but lower margins. the following assumption: If the years preceding the crisis Also, failed banks had a significant relationship between the were highly profitable, a bank would not fail because that percentage of real estate loans and risk. This relationship was profitability could be used to absorb future losses; in other negative in successful banks. Their findings confirmed that words, it could handle the risk. successful banks allocated real estate loans that were both The research model, which is depicted in Figure 1, com- fewer in number and higher in quality. I see this study as a bines five latent variables—profitability of 2006 (Prof ), good example of a causal analysis (i.e., symptoms and risk of 2006 (Risk ), profitability of 2007 (Prof ), risk of 2006 2007 causes), but the measurement invariance of the composite 2007 (Risk ), and profitability of 2008 (Prof )—as 2007 2008 model (MICOM)—which is a critical step in cross-group reflective constructs. For each reflective construct variable, investigation—has never been observed in this study. Without three manifests (indicators) were assigned. Profitability and measurement invariance, a study is susceptible to potential risk of 2007 mediate the path of profitability and risk of 2006 misspecification bias. Therefore, this article differs from the and profitability of 2008. Serrano-Cinca study in that the measurement invariance is Figure 1 presents both direct and indirect (mediating) tested before running a multiple analysis to compare the path effects that can be responsible for a bank’s distress. The prof- coefficients between failed and nonfailed banks. itability and risk in a previous year would affect the profit- To explain the drivers of bank soundness in G7 countries ability of the following year, and so on. The predictive model from 2003 to 2013, Ayadurai and Eskandari (2018) devel- covered the 3 years prior to the banks’ closing in 2009. oped a model with 17 indicators of six constructs as the As mentioned in the introduction, the IPMA focuses direct cause and eight as the indirect cause. They found that mainly on the key target constructs of interest in the PLS banks placed high importance on off-balance sheets and cap- path model. Figure 1 shows that the profitability of 2008 is ital activities, thus taking on higher risk. the final target construct, whereas the predecessor constructs By using PLS-SEM models, Avkiran et al. (2018) moni- are profitability of 2006, profitability of 2007, risk of 2006, tored the transmission of systematic risk from shadow banks and risk of 2007. Starting at the back end of Figure 1, it can to regular banks. The results of the predictive model indi- be seen that banks’ profitability in 2008 was positively or cated that a substantial degree of the variation in systematic negatively influenced by the predecessor constructs. By risk in the regulated banks was explained by micro-level and employing the PLS-SEM using SmartPLS 3 software (Hair macro-level linkages that can be traced to shadow banking. et al., 2016), the PLS model results can be utilized to calcu- In terms of obtaining forecasts for net charge-off rates for late the important scores. The important-performance values banks, J. R. Barth et al. (2018) used a PLS model to extract of the final target construct (i.e., profitability of 2008) were target-specific factors. They included more than 200 predic- created based on the important-performance values of the tor variables utilizing 250 quarterly macroeconomic data predecessor constructs (i.e., profitability of 2006, profitabil- collected from the period 1987: Q1 to 2016: Q4. The empiri- ity of 2007, risk of 2006, and risk of 2007). cal results showed that PLS outperformed benchmark mod- Figure 1 presents several paths (i.e., H1, H2, H3, H4, H5, els. They concluded that their model approach would assist H6, H7, and other indirect paths). In keeping with general banks in determining which variables cause failures and con- IPMA procedures, these path coefficients must be deter- tain losses to manageable levels. mined to be significant at any level of confidence before con- Although these papers presented the advantages of apply- firming that the required conditions for carrying out the ing PLS compared with traditional statistical methods, they IPMA have been established. did not apply the IPMA in this context. The present work, therefore, supplements the literature by reviewing the appli- Data Collection and Analysis Process cation of IPMA as an extension of PLS-SEM and testing the According to the FDIC, 140 American banks failed in 2009 MICOM. because of the financial crisis. To explore the reasons behind this failure and to contribute to the existing body of knowl- Research Method edge, this work analyzes the entire population of failed banks in 2009. This article compares those failed banks (service Path Model Estimation type = 0; n =140) with the nondistressed banks (service type The main goal of banks is to serve customers and obtain = 1; n =140) listed by the FDIC by employing both paramet- profit at the same time. To achieve this objective, banks must ric and nonparametric tests from 2006 to 2008. The total balance risk and return (J. R. Barth et al., 2018). Handling number of observations is 12,600 (280 firms × 3 years × 15 4 SAGE Open Figure 1. Direct and indirect (mediating) effects. indicators). According to Hair et al. (2016), the rule of thumb The measurement model of this study has five latent con- requires the sample size must be at least 10 times the maxi- structs, as depicted in Figure 1. Results of reliability analy- mum number of indicators associated with an outer model, ses using composite values show that all of these constructs which is met in this study. (profitability and risk) obtained a high reliability value above The financial data of this study were collected from the 0.7 thresholds (China et al., 2012). Table 1 shows that all FDIC and are available to the public. For each bank, 15 finan- indicators’ loadings are greater than 0.7 for both groups cial ratios were selected to build the model. Procedures were except for three indictors (Nimy for failed banks, and followed to evaluate the predictive model using PLS-SEM in Nimy and Nimy for nonfailed banks). As evidenced, 2007 2008 SmartPLS 3 software (Hair et al., 2016). Accordingly, the the low loading value of those three indicators ranges from financial ratios were filtered by eliminating those with a weak 0.5 to ≈ 0.7. In exploratory studies, loadings higher than 0.4 effect and using only the ratios with a strong effect to predict are acceptable (Avkiran et al., 2015; Hair et al., 2013, p. 6). failure. According to the rules of thumb of model evaluation This confirms that the internal consistency requirements using PLS-SEM, only six ratios, as defined in Table A1, were were established. employed for each year. Net interest margin (Nimy), return To validate the measurement model, two types of validity on assets (ROA), and return on equity (ROE) are proxy for were checked: convergent validity and discriminant validity. profitability, whereas core capital (leverage) ratio (Rbc1aaj), The average variance extracted (AVE) was used as a crite- Tier 1 risk-based capital ratio (Rbc1rwaj), and total risk-based rion of convergent validity, whereas the Fornell–Larcker cri- capital ratio of risk (Rbcrwaj) are proxy for the capital terion and cross-loadings were utilized as criteria of strength of the bank. To evaluate the univariate discrimina- discriminant validity. tory power of each ratio, a mean test was computed. In establishing convergent validity, a threshold of accept- ability of greater than 0.5 was ensured for the value of AVE. (China et al., 2012; Hair et al., 2016; Rasoolimanesh et al., Model Assessment Utilizing PLS-SEM 2017). Findings presented in Column 4 of Table 1 indicate Before reporting the findings of this work, reliability and that all AVE values met the requirements for the construct validity must be assessed as a prerequisite for accurate esti- measures’ convergent validity, which is above the accept- mation in SEM (Nitzl, 2016). Although previous works offer ability value. The convergent validity ranges from 0.625 to criteria to assess partial model structures, this article follows 0.969. The results showed that the loading’s value of the guideline of Hair et al. (2016). To prevent a common mis- Nimy , Nimy , and Nimy ranged from 0.4 to 0.7, 2006 2007 2008 take with model assessment, the reliable and valid measure- validating the decision to retain those indicators in the model. ment of reflective indicators should not be applied to Finally, the article measures discriminant validity by formative indicators (L. Lee et al., 2011). looking at the square root of AVE for each latent variable by Tailab 5 Table 1. Evaluation Results of the Measurement Model. 2 3 4 Loading Composite reliability AVE 1 Failed Nonfailed Failed Nonfailed Failed Nonfailed Prof 0.880 0.908 0.713 0.768 Nimy 0.673 0.792 ROA 0.882 0.908 ROE 0.953 0.923 Risk 0.977 0.971 0.934 0.917 Rbc1aaj 0.931 0.910 Rbc1rwaj 0.989 0.984 Rbcrwaj 0.978 0.976 Prof 0.901 0.851 0.754 0.669 Nimy 0.777 0.523 ROA 0.906 0.936 ROE 0.914 0.926 Risk 0.988 0.965 0.965 0.901 Rbc1aaj 0.969 0.896 Rbc1rwaj 0.989 0.978 Rbcrwaj 0.989 0.971 Prof 0.914 0.827 0.780 0.625 Nimy 0.842 0.531 ROA 0.924 0.891 ROE 0.881 0.894 Note. AVE = average variance extracted; ROA = return on assets; ROE = return on equity. Table 2. Discriminate Validity (Fornell–Larcker Criterion). Failed banks Nonfailed banks Constructs Prof Prof Prof Risk Risk Prof Prof Prof Risk Risk 2006 2007 2008 2006 2007 2006 2007 2008 2006 2007 Prof .844 .876 Prof −.239 .868 .875 .818 Prof −.297 .337 .883 .563 .725 .790 Risk .103 −.174 −.254 .966 −.575 −.430 −.161 .957 Risk −.140 .414 .189 .018 .982 −.451 −.331 −.119 .913 .949 Note. The values in bold are the square root of AVE, and below the diagonal are the construct correlations. AVE = average variance extracted. applying Fornell–Larcker criterion. Table 2 shows that the discriminant validity is established (Henseler et al., 2015, square root of AVE for each latent variable (in bold) is higher 2016); however, if the value of the HTMT is greater than this than the absolute values of other correlation values among threshold, there is a lack of discriminant validity (Ngah et al., the latent variables. This indicates that all of the discriminant 2018). Table 3 shows that all HTMT values were lower than validity requirements through the square root of AVE to the the threshold value. correlation, and through the cross-loading analysis, are met. In addition to the validity evaluation, this work checks for Although the Fornell–Larcker criterion and cross- multicollinearity by variance inflation factor (VIF) and predic- loadings are widely used for evaluating discriminant validity, tion relevance (Q ). A VIF of greater than 5 suggests a multi- SmartPLS provides yet another reliable criterion called the collinearity issue, and a VIF of greater than 10 reveals a critical heterotrait-monotrait ratio (HTMT). In using the HTMT cri- multicollinearity problem. For stricter tests, VIF should be terion to evaluate discriminant validity, one must determine below 3.3 in the inner model between latent variable predictors the HTMT value for all pairs of reflective constructs. If the (Cenfetelli & Geneviève, 2009, p. 694; Kock & Lynn, 2012; value of HTMT is below .85 (the threshold value), then Lowry & Gaskin, 2014, p. 137). Table 4 presents the outcome 6 SAGE Open Table 3. Discriminate Validity (HTMT Criterion). Constructs Prof Prof Prof Risk Risk 2006 2007 2008 2006 2007 Prof Prof .275 Prof .102 .574 Risk .122 .094 .105 Risk .139 .294 .179 .637 Note. All findings satisfy the HTMT.85 criterion. HTMT = heterotrait-monotrait ratio. 2 2 Table 4. Measurement of Interaction Through Effect Size f and Q . Prof Risk Prof Q² 2007 2007 2008 2 2 2 VIF f VIF f VIF Total sample Failed Nonfailed Constructs f Prof .056 1.013 .004 1.105 Prof .191 1.268 .035 .045 .489 Risk .002 1.013 .636 1.000 .016 1.801 Risk .014 1.966 .349 −.001 .724 Prof .159 .138 .305 SRMR .08 Note. The results of profitability and risk of year (t = 2006) as exogenous latent variables on profitability and risk of year (t + 1 = 2007) and (t + 2 = 2008) as endogenous variables, and the results of profitability and risk of year (t + 1 = 2007) as exogenous variables on profitability of year (t + 2 = 2008) as endogenous variables. VIF = variance inflation factor; SRMR = standardized root mean square residual. of a multicollinearity test for a structural model (inner model) To measure the magnitude of R values as a criterion for that indicates collinearity is not a concern. However, five predictive relevance, this article tests Stone and Geisser’s Q reflective indicators’ values (outer model) exceed the thresh- to be sure that the model has predictive relevance. The Q old of 5. So, multicollinearity arising among these indicators was calculated based on the blindfolding procedure. A posi- has a small correlation with the construct latent variables. tive Q value gives evidence that the path model has predic- Collinearity among these indicators occurs due to the type of tive relevance, whereas a negative value signifies a lack of financial data. In particular, financial ratios that share a com- predictive relevance (Henseler et al., 2009, p. 303). The mon numerator or denominator can suffer from multicol- result demonstrated in Table 5 indicates that all endogenous linearity (Serrano-Cinca & Gutiérrez-Nieto, 2013). We constructs were above zero (positive), which confirms that expected that some degree of collinearity would occur because the path model has predictive relevance. By contrast, Risk the corresponding numerators and denominators are likely to for failed banks has no predictive relevance as Q stands at a be correlated. negative value. R was used to evaluate predictive power. Results indicate that the value of R was considered weak for the whole sam- Measurement Invariance and Multi-Group 2 2 ple (R = .23) and for failed banks (R = .20), whereas it was Analysis considered moderate for nonfailed banks (R = .56) (Chin, 1998). In this endeavor, however, Henseler et al. (2009) Measurement invariance, which is also known as measure- accept moderate R if the model has only one or two exoge- ment equivalence, is a crucial step in cross-group investiga- nous latent variables. It is very difficult to find accepted rules tion (Ruzzier et al., 2014). This technique enables researchers for preferred R because it is usually dependent on the degree to identify whether parameters of the structural model and of model complexity and research discipline (Hair et al., measurement model are equivalent (i.e., invariance) across 2016, p. 199). two or more groups (China et al., 2012). Measurement invari- Unlike R , which looks at each endogenous latent vari- ance is needed to be sure that the differences are not attribut- able, f measures whether the effect of a particular exogenous able to measurement model differences across the groups variable on an endogenous variable is substantive when that (Kock, 2017). If one failed to establish invariance, it would be exogenous variable is omitted from the model. This test is difficult to determine whether the differences observed were called the effect size. According to the Hair et al. (2016, due to true differences (China et al., 2012, p. 1). p. 201) guideline, f values of .02, .15, and .35, respectively, Before running an multiple-group analysis (MGA) to represent small, medium, and large. compare the path coefficients between failed and nonfailed Tailab 7 Table 5. Results of Invariance Measurement Testing Using Permutation. Compositional invariance Configural Partial Equal mean value Equal variance Full invariance (same measurement measurement algorithms for 5% quantile invariance invariance Constructs both groups) C = 1 of Cu p Values established Difference CIs Difference CIs established Prof Yes .996 [.921, 1.000] .684 Yes −0.544 [−.202, .202] 0.157** [−0.331, 0.325] No/Yes Prof Yes .998 [.983, 1.000] .525 Yes −0.682 [−0.204, 0.209] 1.187 [−0.288, 0.295] No/No Prof Yes .999 [.989, 1.000] .500 Yes −0.463 [−0.207, 0.187] 1.506 [−0.313, 0.326] No/No Risk Yes 1.000 [.999, 1.000] .780 Yes −0.795 [−0.201, 0.202] −0.418 [−0.337, 0.321] No/No Risk Yes 1.000 [1.000, 1.000] .300 Yes −0.856 [−0.205, 0.203] −0.305* [−0.330, 0.334] No/Yes Note. If C exceeds the 5% quantile of Cu, compositional invariance is established CI = confidence interval. *p value greater than .05 indicates that the correlation is not significantly lower than 1. **There are no significant differences in the mean and variance values of the latent variables across the two groups because the original difference value is within the corresponding CI. banks, it is important to employ invariance testing to avoid extensive work of Ringle, Sarstedt, and Mooi (2010) and potential misspecification bias and misleading results. The Ringle, Wende, and Will (2010), and implemented in the MICOM test is required in PLS-SEM to be confident that the software application SmartPLS (Sarstedt et al., 2011, p. 35). group differences in the model “do not result from either dis- This approach combines a finite mixture procedure with tinctive content or meanings of the latent variables across an expectation- groups or from the measurement scale” (Sarstedt et al., 2018, maximization (EM) algorithm (Loureiro & Miranda, 2011). p. 209). The MICOM analysis consists of three steps SmartPLS 3 (Hair et al., 2018) was used to segment the sam- (Henseler et al., 2016, p. 412): (a) configural invariance, ple based on the estimated scores for latent variables. which is the first and weakest level of measurement (Kim According to the result of FIMIX-PLS presented in Table 6, et al., 2013, p. 502); (b) compositional invariance; and (c) the the analysis considers two-, three-, and five-segment solu- equality of composite mean values and variances. Thus, if tions. Applying the relevant assessment criteria provided by configural invariance and compositional invariance are Hair et al. (2018), two segments was the appropriate choice. established, partial measurement invariance is confirmed. Because the optimal solution is the number of segments with After that, researchers can compare the path coefficient esti- the lowest value (bold numbers) and the highest value of mates between groups. In addition, if partial measurement entropy values (EN) (Hair et al., 2018, p. 196), Segments 1 invariance is established and the composites have equal and 4 have been eliminated. No single criterion indicates the mean values and variances (Step 3) “across the groups, full same number of segments, whereas the EN range between measurement invariance is confirmed, which supports the 0.683 and 0.831, which are higher than the threshold of 0.5 pooled data analysis” (Henseler et al., 2016, p. 413). Pooled (Hair et al., 2018, p. 197; Schlagel & Sarstedt, 2016, p. 640). data analysis examines whether a common core of relation- Panel B of Table 6 indicates that Segment 5 has very small ships exists across the groups (Durvasula et al., 1993). segment sizes. To warrant valid analysis, Segment 5 was In accordance with the MICOM analysis, Table 5 shows dropped. Panel C of Table 6 confirms that the R value in that partial measurement invariance is established, which is a Segment 2 is the highest compared with the original sample. requirement to compare groups and test any significant It can be concluded that heterogeneity in this study is not differences. prevalent. Robustness Check IPMA To ensure that validity measurement is established when IPMA—also called importance-performance matrix, impact- using PLS-SEM, checking unobserved heterogeneity is very performance map, and priority map analysis (Ringle & important (Hair et al., 2018; Sarstedt et al., 2018; Schlagel & Sarstedt, 2016, p. 1866)—was first proposed and introduced Sarstedt, 2016). Although data of this study are obtained by Martilla and James (1977). The IPMA approach examines from the same population (the banking industry), assumption not only the performance of an item but also the importance of homogeneity is unrealistic because individuals are not of that item. The objective of this analysis is to identify the homogeneous in their perceptions and evaluations of unob- (unstandardized) total effect of predecessor construct’s served constructs (Ansari et al., 2000). This article applies importance (e.g., profitability of 2006) in anticipating a spe- the finite mixture partial least squares (FIMIX-PLS) method, cific target endogenous construct (e.g., profitability 2008) which was proposed by Hahn et al. (2002), advanced by the (Hair et al., 2016, p. 276; Hair et al., 2018, p. 105). The total 8 SAGE Open Table 6. Model Selection. Fit indices Number of segments Criteria 2 3 5 Full data set Panel A: Fit indices for segment solution AIC (Akaike’s information criterion) 1,770.361 1,726.119 1,671.359 AIC3 (modified AIC with Factor 3) 1,791.361 1,758.119 1,725.359 AIC4 (modified AIC with Factor 4) 1,812.361 1,790.119 1,779.359 BIC (Bayesian information criteria) 1,846.691 1,842.432 1,867.638 CAIC (consistent AIC) 1,867.691 1,874.432 1,921.638 HQ (Hannan–Quinn criterion) 1,800.977 1,772.772 1,750.087 MDL5 (minimum description length with Factor 5) 2,320.014 2,563.685 3,084.752 EN (entropy statistic [Normed]) 0.831 0.683 0.733 Panel B: Relative segment sizes Segment sizes, % 0.292* 0.182 0.083 Panel C: FIMIX-PLS, R Prof .924 .160 .810 .058 Prof .924 .504 .975 .233 Note. FIMIX-PLS = finite mixture partial least squares. Figure 2. Importance-performance matrix and path model with results. effect demonstrates the importance of apparent variables, performance (Jaafar et al., 2016; Y.-C. Lee et al., 2008). The whereas the mean value of their scores (ranging from 0, goal is to determine each predecessor construct’s importance which is considered the lowest, to 100, the highest) reflects in terms of its total impact on each target endogenous con- their performance (Höck et al., 2010, p. 201). The interpreta- struct (performance). tion of IPMA is that a 1-unit increase in the predecessor’s Figure 2 shows that the final target construct—profitability performance (e.g., Prof ) increases the performance of the of 2008—was affected both directly and indirectly by the target construct (Prof ) by the size of the predecessor’s profitability and risk of 2006 and 2007. unstandardized total effect (Hair at el., 2016, p. 278). The Starting at the back end of Figure 2, when we look at the IPMA technique has two dimensions: importance and final target construct (i.e., profitability of 2008), profitability Tailab 9 Figure 3. Importance performance map of the profitability of 2008 (failed banks). of 2007 has a relatively high positive importance with the significantly different across the two groups (failed and non- path coefficient of 0.456, whereas profitability of 2006 has failed banks) except for one indicator: Nimy. lower importance with the path coefficient of 0.056. The Descriptive analysis depicted in Table 7 confirms that the path coefficient, depicted as arrows, demonstrates the rela- average profitability was higher for nonfailed than failed tive importance, whereas the performance values, depicted banks. Unlike distressed banks, nonfailed banks had never as circles, are the average values of latent variables’ scores had a negative performance, which protected them from on a scale of 1 to 100. It should be noticed that a score closer bankruptcy. Failed banks had a lower average risk than non- to 100 indicates a higher performance latent variable (Hair failed banks. Hence, low profitability can be a symptom of et al., 2016). failure. The results of IPMA presented in Figure 2 are also depicted in Figures 3 (failed banks) and 4 (nonfailed banks) Test of Significant Paths and discussed more deeply later. A comprehensive under- standing of how to read and use results plotted in those fig- As with any other analytical technique, IPMA needs to be ures can assist management in improving low performance driven by statistical power consideration (Streukens et al., by focusing on high importance (Hock et al., 2010). 2018, p. 380). Primary data analysis confirms interesting The IPMA approach must meet two requirements before findings in six significant paths related to the main objective any application: (a) all indicators must have the same orien- of this article. The results are summarized in Table 8 for each tation, and (b) outer weight must not be negative (Hair et al., subgroup. It is important to consider the sign of the path coef- 2016, p. 208; Hair et al., 2018, p. 123; Ringle & Sarstedt, ficient (positive or negative), as it can be the reason those 2016, p. 1868). These requirements have been met, as shown banks went bankrupt (e.g., negative profit). The path coeffi- in Tables B2 and B3. cients were tested (using 5,000 bootstrap) by applying a one- tailed test (for a more detailed description of when or whether to utilize one-tailed or two-tailed tests, see Kock, 2015). Results The study found that the profitability in a given year was significantly correlated to the profitability of a following Descriptive Statistic year. In failed banks, it was found that the profitability of Table 7 presents the descriptive statistics of our full sample, 2006 (Prof → Prof , β = −0.233, p < .000) was nega- 2006 2007 as well as of subsamples of financial performance. Results tively associated with the profitability of 2007. In other reported in Table 7 confirm that all indicators (ratios) are words, for nonfailed banks, the study observes a strong 10 SAGE Open Table 7. Summary Statistics of the Two Groups (Failed and Nonfailed) and the Whole Sample. Total sample (N = 280) Failed banks (N= 140) Nonfailed banks (N = 140) Univariate test Construct p-value mean variables Indicators M SD M SD M SD test Profitability Nimy 0.038 0.011 0.037 0.01 0.04 0.01 .238 ROA −0.001 0.024 −0.009 0.025 0.01 0.02 (.000) ROE 0.048 0.11 0.003 0.116 0.09 0.08 (.000) Risk Rbc1aaj 0.101 0.043 0.089 0.039 0.11 0.04 (.000) Rbc1rwaj 0.13 0.054 0.107 0.042 0.15 0.05 (.000) Rbcrwaj 0.139 0.048 0.119 0.039 0.16 0.05 (.000) Profitability Nimy 0.038 0.011 0.037 0.012 0.04 0.01 .222 ROA −0.003 0.027 −0.014 0.033 0.01 0.01 (.000) ROE 0.04 0.107 −0.001 0.117 0.08 0.08 (.000) Risk Rbc1aaj 0.096 0.039 0.082 0.036 0.11 0.04 (.000) Rbc1rwaj 0.125 0.052 0.101 0.041 0.15 0.05 (.000) Rbcrwaj 0.134 0.048 0.113 0.038 0.16 0.05 (.000) Profitability Nimy 0.037 0.011 0.037 0.014 0.04 0.01 .366 ROA −0.005 0.029 −0.014 0.037 0 0.01 (.000) ROE 0.028 0.107 0.004 0.125 0.05 0.08 (.000) Risk Rbc1aaj 0.087 0.039 0.073 0.042 0.1 0.03 (.000) Rbc1rwaj 0.114 0.053 0.089 0.045 0.14 0.05 (.000) Rbcrwaj 0.124 0.051 0.1 0.045 0.15 0.04 (.000) Note. Path coefficient is significant at p < .05 for one-tailed test. ROA = return on assets; ROE = return on equity. Table 8. Results of Hypothesis Testing. Path coefficient CIs bias-corrected Failed Nonfailed Failed Nonfailed Hypotheses Relationship β t value β t value 5.0% 0.95% 5.0% 0.95% Supported HI Prof → Prof −0.233* 2.937 0.938* 20.807 [−0.338, −0.076] [0.858, 1.003] Yes/Yes 2006 2007 H2 Prof → Prof −0.214* 3.099 −0.174 0.893 [−0.323, −0.093] [−0.507, 0.130] Yes/No 2006 2008 H3 Prof → Prof 0.223* 2.662 0.947* 5.987 [0.072, 0.353] [0.690, 1.213] Yes/Yes 2007 2008 H4 Risk → Prof −0.151* 1.683 0.109 1.629 [−0.286, −0.001] [−0.012, 0.208] Yes/No 2006 2007 H5 Risk → Prof −0.194* 2.577 0.243 0.927 [−0.319, −0.071] [−0.134, 0.719] Yes/No 2006 2008 H6 Risk → Risk 0.018 0.178 0.913* 31.860 [−0.125, 0.202] [0.858, 0.955] No/Yes 2006 2007 H7 Risk → Prof 0.070 0.702 −0.106 0.447 [−0.092, 0.237] [−0.528, 0.257] No/No 2007 2008 H8 Prof → Profi → Prof −0.054* 1.797 0.895* 5.321 [−0.100, −0.011] [0.636, 1.187] Yes/Yes 2006 2007 2008 H9 Risk → Prof → Prof −0.034 1.405 0.097 1.561 [−0.082, −0.003] [−0.012, 0.208] No/No 2006 2007 2008 H10 Risk → Risk → Prof 0.001 0.103 −0.097 0.443 [−0.010, 0.035] [−0.490, 0.236] No/No 2006 2007 2008 Note. CI = confidence interval. *p < .05 for one-tailed test because the sign “+” or “−” is important. positive association between the profitability of 2006 the profitability of 2007 for failed banks decreased by (Prof → Prof , β = 0.938, p < .000) and the profit- −0.233, whereas the profitability of nonfailed banks 2006 2007 ability of 2007. The regression coefficient indicates that increased by 0.938. when the profitability of 2006 changed by 1 unit (dollar), Furthermore, in both groups, there was a positive and sig- with the assumption that other factors remained constant, nificant effect of the profitability of 2007 on the profitability Tailab 11 Table 9. Results of Hypothesis Testing (MGA-PLS). Path coefficient Welch– Hypothesis Path differences p-value MGA Parametric test Satterthwaite test Supported H1 Prof → Prof 1.161 1.00** 0.000* 0.000* Yes 2006 2007 H2 Prof → Prof 0.040 0.59 0.847 0.847 No 2006 2008 H3 Prof → Prof 0.723 1.00** 0.000* 0.000* Yes 2007 2008 H4 Risk → Prof 0.261 0.99** 0.018* 0.019* Yes 2006 2007 H5 Risk → Prof 0.437 0.95 0.107 0.108 No 2006 2008 H6 Risk → Risk 0.895 1.00** 0.000* 0.000* Yes 2006 2007 H7 Risk → Prof 0.176 0.26 0.495 0.495 No 2007 2008 H8 Prof → Prof → Prof 0.938 1.00** 0.000* 0.000* Yes 2006 2007 2008 H9 Risk → Prof → Prof 0.039 0.584 0.865 0.865 No 2006 2007 2008 Note. **p value of MGA lower than .05 or higher than .95 indicates significant differences at the 5% level. Parametric and Welch–Satterthwaite tests are significant for. *p value only at the 5% level, which is lower than .05. PLS-SEM = partial least squares structural equation modeling; MGA = multiple- group analysis. This indicates the difference between the group-specific path coefficients, which is not necessary to be lower than 1 (Christian Ringle, Personal communication, April 17, 2018). of 2008 (Prof → Prof , β = 0.223, p < .000 and β = To measure the difference between the two groups (failed 2007 2008 0.947, p < .000) for failed and nonfailed banks, respectively. and nonfailed banks), the study employed Henseler’s MGA, There was significant evidence that the profitability of 2006 the parametric test, and the Welch–Satterthwaite test. had a negative impact on the performance of 2008 for failed Findings illustrated in Table 9 reveal significant differences banks (Prof → Prof , β = −0.214, p < .000). This was between failed and nonfailed banks in five paths. First, the 2006 2008 not the case for nonfailed banks because the study could not significant difference for Prof → Prof (H1) was 1.161. 2006 2007 find a significant correlation. This pointed to the difference between the group-specific In terms of risk effect, the current study indicated several path coefficients, which need not be lower than 1.0 to estab- significant paths. First, the path coefficient for Prof → lish. For example, if the path coefficient is 0.5 for one group Risk was significantly negative (β = −0.151, p < .000) and −0.6 for another group, then the difference between the for failed banks. The path coefficient for Prof → Risk two groups is 1.1 (Christian Ringle, personal communica- 2008 2006 was also significantly negative (β = −0.194, p < .000). tion, April 17, 2018). Second, the path coefficient for Prof These paths were not significant for nonfailed banks. This → Prof (H3) differed between the two groups by 0.723, result suggested that for failed banks when the risk changed which was significant according to the p value of Henseler’s by 1 unit, it tended to decrease the profitability. This was not MGA and permutation and Welch–Satterthwaite tests. Third, the case for nonfailed banks. Therefore, it can be assumed the path coefficient for Prof → Risk (H4) supported a 2007 2006 that this was the potential reason behind failure. However, significant difference of 0.261. In addition, the path coeffi- the path coefficient for Risk → Risk was significantly cient for Risk → Risk (H6) revealed a significant dif- 2006 2007 2007 2008 positive (β = 0.913, p < .000) for nonfailed banks (only). ference between the two groups of 0.895. Finally, the results Finally, the study revealed no significant coefficient for indicated that one of the two mediating paths was signifi- Prof → Risk for both groups. Thus, the effect of cantly different. The indirect path Prof → Prof → 2008 2007 2006 2007 Risk on Prof was unsupported in both groups. Prof (H8) differed by 0.938. The results did not support a 2007 2008 2008 The indirect (mediating) effect’s findings illustrated that significant difference with regard to the effects of Prof → one of the three mediating paths was significant. Findings Prof (H2), Risk → Prof (H5), or Risk → Prof 2008 2006 2008 2006 2007 illustrated in Table 8 show that the indirect path for Prof → Prof (H9). 2006 2008 → Prof → Prof was significantly negative for failed 2007 2008 banks (β = −0.054, p < .000), whereas it was highly posi- Results of IPMA. With bootstrapping with 5,000 subsamples tive for nonfailed banks (β = 0.895, p < .000). However, confirming that some of the path coefficients were statisti- other specific indirect effects were not significant. This cally significant and the requirements for carrying out the result indicated that consecutive low profitability of failed IPMA approach have been met, the findings of IPMA can banks caused the failure, unlike the consecutive high profit- finally be discussed. These results are plotted in Figure 3 ability of nonfailed banks that was used to absorb future (i.e., failed banks) and Figure 4 (i.e., nonfailed banks). In losses. This result reflected the difference in strategy each importance-performance map, the analysis concen- between failed and nonfailed banks. The study observed no trated on the lower right area to enhance improvement significant mediating role of Prof and Risk between because items plotted in that area have high importance with 2007 2007 Risk and Prof . low performance. Concentrating constructive action in this 2006 2008 12 SAGE Open Figure 4. Importance performance map of the profitability of 2008 (nonfailed banks). area will produce maximum results (Martilla & James, 1977, Although Figure 4 presents that the profitability of 2006 p. 78). For failed banks, the results confirmed that the profit- was the second-highest importance indicator, it was not a ability of 2007 and risk of 2007 had particularly high impact significant important variable in the projection of the perfor- on the profitability of 2008. mance of 2008. It is evident from Figure 3 that the profitability of 2007 had a large impact on the profitability of 2008 in failed banks, Discussion and Managerial Implications and thus represents a major opportunity for improvement that could have been addressed by bankers’ activities. More The findings suggest that there is a significant difference precisely, the importance of the profitability of 2007 had a between failed and nonfailed banks regarding the effect of positive total effect on 2008: a 1-dollar increase in the profit- the financial crisis in 2007 and 2008. Research on banks’ ability of 2007 increased the profitability of 2008 by 0.254, performance during the crisis (Beltratti & Stulz, 2012) whereas changing the risk by 1 dollar in 2007 tended to showed that banks were affected differentially because they increase the profitability of 2008 by 0.047. This indicated had different balance sheets and profitability before the cri- that there was indeed room to improve the performance of sis. For failed banks, negative profitability measured by 2008. As seen in Figure 3, the profitability of 2006 had the ROA and a fall in ROE appeared 3 years before failure. strongest negative total effect on the profitability of 2008. Bankers should have dealt with this trend as a symptom of This variable was very important to affect the performance failure. This result is consistent with Beltratti and Stulz of 2008 because the total effect was significant. This study (2012), who confirmed that bank profitability in 2006 played found that the total indirect effect of the profitability of 2006 a more significant role in determining bank performance dur- on the performance of 2007 through 2008 was negative ing the crisis than did other factors such as bank governance (–0.056). Both direct and indirect total effects confirmed that and bank regulation. This was not apparent in the case of banks failed in 2008 because of their financial performance nonfailed banks whose profitability was never negative. For in 2006. nonfailed banks, increasing the income tended to increase It is evident from Figure 4 that for nonfailed banks, the the return, which was reflected in profitability ratios as a highest performance indicator was the profitability of 2007. consequence of healthy growth. This result is in line with the This factor was confirmed to be a significant predictor of the accounting theory that assumes that firms can use their net profitability of 2008. Unlike the failed banks, IPMA showed income to absorb future losses. that a 1-dollar increase in the profitability of 2007 increased Turning to performance and risk-taking, it can be seen the profitability of 2008 by 0.887 cents. that the risk that was taken in 2006 led to the poor Tailab 13 performance of failed banks over the next 2 years. However, the performance of 2007 by using their net income to either the risk that was taken by nonfailed banks in 2006 had no recover their previous losses or absorb future losses. correlation with bank outcomes. This is a striking result—it In particular, this study underscores that the most detri- was observed that the average risk (e.g., core capital (lever- mental managerial decisions and policy activities by failed age) ratio, Tier 1 risk-based capital ratio, and total risk-based banks were in regard to high-risk loans. The study indicates capital ratio) accepted by failed banks was lower than the that the risk that was taken in 2007 should have received average risk accepted by nonfailed banks. That lower aver- significant attention because it had a positive total effect on age was the cause of U.S. bank failures during the financial the profitability of 2008 and represented an opportunity for crisis. This finding agreed with the study by Beltratti and failed banks to increase profitability in 2008. Unfortunately, Stulz (2012), who found that large banks with more Tier 1 bankers missed this remarkable opportunity to avert disaster capital and more deposit financing at the end of 2006 had before the crisis arrived. In consequence, loans went unpaid, significantly higher returns during the crisis. This evidence is profitability went down, and banks closed. Applying IPMA also consistent with Serrano-Cinca et al. (2014) who reported allows companies to determine the best allocation of their that nonfailed banks compensated for increases in risk by resources based on the results of the IPMA. The interpreta- strengthening their core capital. However, Cox et al. (2017) tion of the comparison results between failed and nonfailed found that banks failed during the 2008–2010 financial crisis banks was satisfactory because test measurement invariance because they accepted more risk, specifically by having was done to avoid potential misspecification bias. higher financial leverage. This is in line with Serrano-Cinca In closing, managers who use and apply IPMA will et al. (2014), who showed that 5 years before the crisis, failed obtain useful conceptual insights by overlaying the impor- banks had higher loan growths, higher concentration on real tance-performance analysis to prioritize their financial estate loans, higher risk ratios, and higher turnover, but lower decision-making. margins. It can be concluded that the risk that was taken a few years before the crisis had a significant effect on a bank’s Conclusion performance. It remains to be determined why risk taken in 2007 was not significantly associated with profitability in Overall, this research addresses an overview of the applica- 2008. tion of the IPMA as a useful technique to expand the analysis This article extends the application of PLS-SEM by using of PLS-SEM results. This work develops a model to help IPMA to determine priority factors and should be useful to researchers who are keen on applying IPMA in the banking managers seeking to improve banking performance (espe- and finance fields. The results of IPMA indicated that failed cially managers of those banks that failed). The predictive banks were predisposed to decreasing financial performance model developed for this study indicated that failed banks in 2008 because of poor performance in 2006. Conversely, were predisposed to decreasing financial performance in the profitability of 2006 and 2007 positioned nonfailed banks 2008 because of poor performance in 2006. Conversely, the for increasing financial performance in 2008. It must be profitability of 2006 and 2007 positioned nonfailed banks for borne in mind that this research was only conducted on a increasing financial performance in 2008. For failed banks, small group of failed and nonfailed U.S. banks over a short financial performance in 2006 was found to be the highest period. Further research is thus needed to investigate the per- negative important factor that bankers should have addressed formance of large banks around the world during the crisis to avoid failure. Failed banks should have taken advantage of before generalized conclusions can be drawn. 14 SAGE Open Appendix A Table A1. Definition of Variables. Portability Nimy Net interest margin Total interest income minus the total interest expense as a percentage of average earning assets ROA Return on assets Net income after taxes and extraordinary items as a percentage of average total assets. ROE Return on Equity Annualized net income as a percentage of average equity on a consolidated basis. Risk Rbc1aaj Core capital (leverage) ratio Tier 1 (core) capital as a percentage of average total assets minus ineligible intangibles Rbc1rwaj Tier 1 risk-based capital ratio Tier 1 (core) capital as a percentage of risk-weighted assets as defined by the appropriate federal regulator for prompt corrective action during that time period Rbcrwaj Total risk-based capital ratio Total risk-based capital as a percentage of risk-weighted assets Note. ROA = return on assets; ROE = return on equity. Appendix B Table B1. Unstandardized Outer Weights. Prof Risk Prof Risk Profit 2006 2006 2007 2007 2008 Indictor Failed Nonfailed Failed Nonfailed Failed Nonfailed Failed Nonfailed Failed Nonfailed Nimy 0.28 0.36 ROA 0.32 0.37 ROE 0.55 0.41 Rbc1aaj 0.32 0.33 Rbc1rwaj 0.37 0.36 Rbcrwaj 0.35 0.36 Nimy 0.34 0.28 ROA 0.43 0.47 ROE 0.38 0.45 Rbc1aaj 0.36 0.34 Rbc1rwaj 0.33 0.36 Rbcrwaj 0.33 0.35 Nimy 0.35 0.34 ROE 0.42 0.45 ROA 0.35 0.46 Note. ROA = return on assets; ROE = return on equity. Tailab 15 Table B2. Rescaled. Prof Risk Prof Risk Prof 2006 2006 2007 2007 2008 Failed Nonfailed Failed Nonfailed Failed Nonfailed Failed Nonfailed Failed Nonfailed Nimy 0.61 0.57 ROA 0.28 0.34 ROE 0.11 0.09 Rbc1aaj 0.31 0.34 Rbc1rwaj 0.34 0.31 Rbcrwaj 0.35 0.35 Nimy 0.64 0.41 ROA2007 0.28 0.51 ROE 0.07 0.08 Rbc1aaj 0.37 0.39 Rbc1rwaj 0.31 0.29 Rbcrwaj 0.32 0.32 Nimy 0.64 0.51 ROA 0.29 0.41 ROE 0.07 0.08 Note. ROA = return on assets; ROE = return on equity. 6. In accordance with the basic PLS algorithm procedure, the Acknowledgments results presented in this section were obtained after Iteration I would like to thank Mike Tsionas (editor), four anonymous refer- 13; the bootstrapping procedure with 5,000 resamples with ees, and Themistoclis Pantos for their insightful, supportive feed- one-tailed t test and the blindfolding procedure with an omis- back, and comments on this research paper. I also thank Ms. Ashwini sion distance of 7 to Prof as the final target constructs were Thakur from Lincoln University for her research assistance. applied. 7. These criteria are generally consistent with formative latent Declaration of Conflicting Interests variable theory (for more details, see Kock, 2014, p. 11). The author(s) declared no potential conflicts of interest with respect 8. Using the value of R to evaluate model fit might allow to the research, authorship, and/or publication of this article. researchers to describe a model with good fit as poor. Thus, it can be said that fitness measures should only be based on how well the parameters are able to match the sample covariance Funding (Chin, 2010, p. 657). The author(s) disclosed receipt of the following financial support 9. It should be noted that a low value represents a negative out- for the research, authorship, and/or publication of this article: This come, whereas a high value represents a positive outcome. work was supported by the President’s Faculty Grant Program, However, it cannot be concluded that higher latent variables Lincoln University. indicate better performance (Hair et al., 2016, p. 280). ORCID iD References Mohamed M. Khalifa Tailab https://orcid.org/0000-0001- Adebambo, B., Brockman, P., & Yan, X. (2015). Anticipating the 6521-7487 2007–2008 financial crisis: Who knew what and when did they know it? Journal of Financial and Quantitative Analysis, Notes 50(4), 647–669. 1. This is the main reason why this article adopts partial least Ansari, A., Jedidi, K., & Jagpal, S. (2000). A hierarchical Bayesian squares (PLS) instead of an econometric fashion. methodology for treating heterogeneity in structural equa- 2. As one of the general importance-performance map analysis tion models. Marketing Science, 19(4), 328–347. https://doi. (IPMA) procedures, these path coefficients must be signifi- org/10.1287/mksc.19.4.328.11789 cant at any level of confidence before checking whether the Avkiran, N. K. (2018). Rise of the partial least squares struc- requirements for carrying out the IPMA have been established. tural equation modeling: An application in banking. In N. K. 3. According to Ringle and Sarstedt (2016, p. 1882), an indica- Avkiran & C. M. Ringle (Eds.), Partial least squares structural tor-related analysis is particularly useful in formative measure- equation modeling: Recent advances in banking and finance ment model settings. This work neglected the indicator-related (international series in operations research & management analysis because its indicators are reflective instead. science) (p. 239). Springer. 4. The previous work will be presented in this section due to its Avkiran, N. K., Ringle, C., & Low, R. (2015, July 16). Monitoring chronology. transmission of systemic risk from shadow banking to regulated 5. These requirements will be discussed later in this work (see banking. http://papers.ssrn.com/sol3/papers.cfm?abstract_id= section “IPMA.” 2634184 16 SAGE Open Avkiran, N. K., Ringle, C. M., & Low, R. (2018). Monitoring trans- Range Planning, 46(1–2), 1–12. doi:https://doi.org/10.1016/j. mission of systemic risk: Application of partial least squares lrp.2013.01.001 structural equation modeling in financial stress testing. Journal Hair, J. F., Sarstedt, M., & Ringle, R. M. (2018). Advanced issues in of Risk, 20(5), 83–115. partial least squares structural equation Modeling. Sage. Ayadurai, C., & Eskandari, R. (2018). Bank soundness: A PLS- Henseler, J., Ringle, C. M., & Sarstedt, M. (2015). A new criterion for SEM approach. In N. I. Avkiran & C. M. Ringle (Eds.), Partial assessing discriminant validity in variance-based structural equa- least squares structural equation modeling: Recent advances tion modeling. Journal of the Academy of Marketing Science, in banking and finance (pp. 31–52). Springer. 43(1), 115–135. https://doi.org/10.1007/s11747-014-0403-8 Barth, J. R., Joo, S., Kim, H., Lee, K. B., Maglic, S., & Shen, X. Henseler, J., Ringle, C. M., & Sarstedt, M. (2016). Testing mea- (2018). Forecasting net charge-off rates of banks: A PLS surement invariance of composites using partial least squares. approach. Department of Economics. http://cla.auburn.edu/ International Marketing Review, 33(3), 405–431. econwp/Archives/2018/2018-03.pdf Henseler, J., Ringle, C. M., & Sinkovics, R. R. (2009). The use Barth, M. E., & Landsman, W. R. (2010). How did financial report- of partial least squares path modeling in international market- ing contribute to the financial crisis? European Accounting ing. Advances in International Marketing, 20, 277–319. http:// Review, 19(3), 399–423. dx.doi.org/10.1108/S1474-7979(2009)0000020014 Beltratti, A., & Stulz, R. (2012). The credit crisis around the globe: Höck, C., Ringle, C. M., & Sarstedt, M. (2010). Management of Why did some banks perform better? Journal of Financial multi-purpose stadiums: Importance and performance measure- Economics, 105(1), 1–17. ment of service interfaces. International Journal of Services Cenfetelli, R. T., & Geneviève, B. (2009). Interpretation of Technology and Management, 14(2–3), 188–207. Formative Measurement in Information Systems Research. Jaafar, N. A., Noor, Z. M., & Mohamed, M. (2016). Student rat- MIS Quarterly, 33(4), 689–708. ings of teaching effectiveness: An Importance—Performance Chin, W. W. (1998). The partial least squares approach for structural Analysis (IPA). Journal of Educational and Social Research, equation modeling. In G. A. Marcoulides (Ed.), Methodology 6(3), 33–44. for business and management. Modern methods for business Joseph, M., Allbright, D., Stone, G., Sekhon, Y., & Tinson, J. (2005). research (pp. 295–336). Lawrence Erlbaum. Importance-performance analysis of UK and US bank customer Chin, W. W. (2010). How to write up and report PLS analyses. perceptions of service delivery technologies. International In W. W. Chin (Ed.), Handbook of partial least squares: Journal of Financial Services Management, 1(1), 66–88. Concepts, methods and applications (pp. 665–690). Springer. Kim, J., Chung, H., Amtmann, D., & Revicki, D. A. (2013). China, W. W., Mills, A. M., Steel, D. J., & Schwarz, A. (2012, May Measurement invariance of the PROMIS pain interference item 19-May 22). Multi-group invariance testing: An illustrative bank across community and clinical samples. Quality of Life comparison of PLS permutation and covariance-based SEM Research, 22, 502–507. http://dx.doi.org/10.1007/s11136-012- invariance analysis [Conference session] 7th International 0191-x Conference on Partial Least Squares and Related Methods, Kock, N. (2014). Advanced mediating effects tests, multi-group Houston, United States. analyses, and measurement model assessments in PLS-based Cox, R. A., Kimmel, R. K., & Wang, G. W. (2017). Drivers of US SEM. International Journal of e-Collaboration, 10(1), 1–13. bank failures during the financial crisis. International Journal Kock, N. (2015). One-tailed or two-tailed P values in PLS-SEM? of Business & Management, 12(8). https://doi.org/10.5539/ International Journal of e-Collaboration, 2(1), 1–7. ijbm.v12n8p19 Kock, N. (2017, June 2). Explore Measurement Invariance in Durvasula, S., Andrews, C. J., Lysonski, S., & Netemeyer, R. G. WarpPLS. WarpPLS. https://www.youtube.com/watch?v=29V (1993). Assessing the cross-national applicability of consumer qsAjhzqQ&feature=youtu.be behavior models: A model of attitude toward advertising in Kock, N., & Lynn, G. S. (2012). Lateral collinearity and misleading general. Journal of Consumer Research, 19(4), 626–636. results in variance-based SEM: An illustration and recommen- https://doi.org/10.1086/209327 dations. Journal of the Association for Information Systems, Erkens, D. H., Hung, M., & Matos, P. (2012, January). Corporate 13(7), 546–580. governance in the 2007–2008 financial crisis: Evidence Lee, L., Petter, S., Fayard, D., & Robinson, S. (2011). On the use from financial institutions worldwide. Journal of Corporate of partial least squares path modeling in accounting research. Finance, 18, 389–411. International Journal of Accounting Information Systems, Georgantopoulos, A. G., & Tsamis, A. D. (2013). Assessing the 14(2), 305–328. http://dx.doi.org/10.1016/j.accinf.2011.05.002 efficiency of commercial banks in Greece during the financial Lee, Y.-C., Yen, T.-M., & Tsai, C.-H. (2008). Using importance- crisis: A linear approach in conjunction with financial analysis. performance analysis and decision making trial and evaluation Journal of Money, Investment and Banking, 28, 31–46. laboratory to enhance order-winner criteria ~ a study of com- Hahn, C., Johnson, M. D., Herrmann, A., & Huber, F. (2002). puter industry. Information Technology Journal, 7(3), 396–408. Capturing customer heterogeneity using a finite mixture PLS Loureiro, S. M., & Miranda, F. J. (2011). Brand equity and brand loy- approach. Schmalenbach Business Review, 54, 243–269. alty in the Internet banking context: FIMIX-PLS market segmen- Hair, J. F., Hult, G. T., Ringle, C. M., & Sarstedt, M. (2016). A tation. Journal of Service Science and Management, 4, 476–485. primer on Partial Least Squares Structural Equation Modeling Lowry, P. B., & Gaskin, J. (2014). Partial Least Squares (PLS) (PLS-SEM) (2nd ed.). Sage. Structural Equation Modeling (SEM) for building and testing Hair, J. F., Ringle, C. M., & Sarstedt, M. (2013, February–April). behavioral causal theory: When to choose it and how to use Partial least squares structural equation modeling: Rigorous it. IEEE Transactions on Professional Communication, 57(2), applications, better results and higher acceptance. Long 123–146. Tailab 17 Maricica, M., & Georgeta, V. (2012). Business failure risk Sarstedt, M., Becker, J.-M., Ringle, C. M., & Schwaiger, M. analysis using financial ratios. Procedia – Social and (2011). Uncovering and treating unobserved heterogeneity Behavioral Sciences, 62, 728–732. https://doi.org/10.1016/j. with FIMIX-PLS: Which model selection criterion provides sbspro.2012.09.123 an appropriate number of segments? Schmalenbach Business Martilla, J. A., & James, J. C. (1977). Importance-performance Review, 63(1), 34–62. analysis. Journal of Marketing, 41(1), 77–79. Sarstedt, M., Ringle, C. M., & Hair, J. F. (2018). Treating unob- Ngah, R., Ramayah, T., & Sarmidy, R. (2018). Partial least square served heterogeneity in PLS-SEM: A multi-method approach. analysis on micro enterprises’ intellectual capital and per- In H. Latan & R. Noonan (Eds.), Partial least squares path formance: The mediating effect of tacit knowledge sharing. modeling: Basic concepts, methodological Issues and Journal of Applied Structural Equation Modeling, 2(2), 22–33. Applications (pp. 197–213). Springer. Nitzl, C. (2016). The use of partial least squares structural equa- Schlagel, C., & Sarstedt, M. (2016). Assessing the measurement tion modelling (PLS-SEM) in management accounting invariance of the four-dimensional cultural intelligence scale research: Directions for future theory development. Journal of across countries: A composite model approach. European Accounting Literature, 37, 19–35. http://dx.doi.org/10.1016/j. Management Journal, 34, 633–649. acclit.2016.09.003 Serrano-Cinca, C., Fuertes-Callén, Y., Gutiérrez-Nieto, B., & Ramayah, T., Chiun, L., Rouibah, K., & May, O. (2014). Cuellar-Fernández, B. (2014). Path modelling to bank- Identifying priority using an Importance-Performance Matrix ruptcy: Causes and symptoms of the banking crisis. Applied Analysis (IPMA): The case of Internet banking in Malaysia. Economics, 46(31), 3798–3811. International Journal of E-Adoption, 6, 1–15. https://doi. Serrano-Cinca, C., & Gutiérrez-Nieto, B. (2013). Partial Least org/10.4018/ijea.2014010101 Square Discriminant Analysis for bankruptcy prediction. Rasoolimanesh, S. M., Ringle, C. M., Jaafar, M., & Ramayah, T. Decision Support Systems, 54, 1245–1255. (2017). Urban vs. rural destinations: Residents’ perceptions, Silva, F., & Fernandes, F.O. (2011). Importance-Performance community participation and support for tourism development. Analysis as a Tool in Evaluating Higher Education Service Tourism Management, 60, 147–158. https://doi.org/10.1016/j. Quality: The Empirical Results of ESTiG (IPB). Creating tourman.2016.11.019 Global Competitive Economies: A 360-Degree Approach Ringle, C. M., & Sarstedt, M. (2016). Gain more insight from your Streukens, S., Leroi-Werelds, S., & Willems, K. (2018). Dealing PLS-SEM results The importance-performance map analysis. with nonlinearity in Importance-Performance Map Analysis Industrial Management & Data Systems, 116(9), 1865–1886. (IPMA): An integrative framework in a PLS-SEM context. In https://doi.org/10.1108/IMDS-10-2015-0449 H. Latan & R. Noonan (Eds.), Partial least squares path mod- Ringle, C. M., Sarstedt, M., & Mooi, E. A. (2010). Response- eling: Basic concepts, methodological issues and applications based segmentation using finite mixture partial least squares: (p. 452), Springer. Theoretical foundations and an application to American cus- Wyród-Wróbel, J., & Biesok, G. (2017). Decision making on vari- tomer satisfaction index data. In R. Stahlbock, S. F. Crone, & ous approaches to Importance-Performance Analysis (IPA). S. Lessmann (Eds.), Data mining: Annals of information sys- Faculty of Business and Economics, 3(2), 123–131. https://doi. tems (vol. 8, pp. 19–49). New York: Springer. org/10.11118/ejobsat.v3i2.82 Ringle, C. M., Wende, S., Will, A. (2010). Finite mixture partial Zaghdoudi, T. (2013). Bank failure prediction with logistic regres- least squares analysis: Methodology and numerical examples. sion. International Journal of Economics and Financial Issues, In V. Esposito Vinzi, W. Chin, J. Henseler, H. Wang (Eds.), 3(2), 537–543. Handbook of partial least squares. Springer handbooks of com- putational statistics (pp. 195–218). Berlin, Heidelberg: Springer. Author Biography Ruzzier, M. K., Antoncic, B., & Ruzzier, M. (2014). Cross-cultural model of customer-based brand equity for a tourism destina- Mohamed M. Khalifa Tailab is an assistant professor of busi- tion. IUP Journal of Brand Management, XI, 7–29. ness administration and accounting at Lincoln University. His Samar, S., Ghani, M. A., & Alnaser, F. (2017). Predicting cus- primary research interests include corporate disclosure, textual tomer’s intentions to use internet banking: The role of tech- analysis modeling, market reaction, and the application of partial nology acceptance. Management Science Letters, 7(11), least squares structural equation modeling (PLS-SEM) in busi- 513–524. ness research. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png SAGE Open SAGE http://www.deepdyve.com/lp/sage/using-importance-performance-matrix-analysis-to-evaluate-the-financial-cAlL0uRhA7

Loading next page...

References (50)

Publisher: SAGE
Copyright: Copyright © 2022 by SAGE Publications Inc, unless otherwise noted. Manuscript content on this site is licensed under Creative Commons Licenses.
ISSN: 2158-2440
eISSN: 2158-2440
DOI: 10.1177/2158244020902079
Publisher site: See Article on Publisher Site

Abstract

This research applies a technique that identifies areas of improvement that can be addressed by managerial decisions or policy activities. It extends the application of partial least squares structural equation modeling (PLS-SEM) using an importance- performance map analysis (IPMA). The IPMA determines priority factors that should receive management’s attention. The PLS path model was tested by comparing 140 failed U.S. banks with the same number of nonfailed banks from 2006 to 2008. This model assembles 15 indicators with four predecessor constructs (i.e., profitability of 2006, profitability of 2007, risk of 2006, and risk of 2007) and one final target construct (i.e., profitability of 2008). Profitability and risk of 2007 mediate the path of profitability and risk of 2006 and profitability of 2008. The IPMA indicated that failed banks were predisposed to decreasing financial performance in 2008 because of their poor performance in 2006 and 2007. Conversely, nonfailed banks were more likely to experience increasing financial performance in 2008 because of their positive performance in 2006 and 2007. This study indicates that managers who use IPMA to prioritize their financial decisions will obtain useful conceptual insights and are unlikely to be misled. Although IPMA can be conducted on the indicator level as well, this article limits its analysis by focusing on the construct level only. The use of IPMA is ubiquitous in end-user surveys, but its application to banking is still in its embryonic state. For originality, this work prioritizes the application of IPMA using secondary data collected from financial statements to assess the performance of American banks during the crisis. Keywords partial least squares (PLS), structural equation modeling (SEM), importance-performance map analysis (IPMA), banking crisis, measurement invariance, multiple-group analysis performance of banks is very important because they were at Introduction the center of that crisis (M. E. Barth & Landsman, 2010). In Researchers and professionals widely accept that the bank- the wake of the financial crisis, stakeholders are becoming ing sector is the most important component of any financial increasingly concerned with their firm’s financial perfor- system (Georgantopoulos & Tsamis, 2013). Ergo, the stabil- mance; bankers recognize the need to formulate better strate- ity of the banking system contributes to the stability of eco- gies to drive performance but may struggle to determine nomic growth. Thus, the financial crisis in 2007–2008 caused priorities. the biggest economic disruption since the Great Depression Although an unprecedented number of banks collapsed or (Adebambo et al., 2015), shaking the fiscal world with a were bailed out by governments during the crisis (Erkens wave of massive losses (Zaghdoudi, 2013, p. 537). et al., 2012), not all banks across the world performed equally According to the Federal Deposit Insurance Corporation poorly; some banks performed better during the crisis (FDIC), 140 U.S. banks failed in 2009, including several high-profile institutions such as Bear Stearns, Citigroup, Lehman Brothers, Merrill Lynch, and Wachovia. This wide- Lincoln University, Oakland, CA, USA spread failure indicated the weaknesses of the banking system Corresponding Author: (Ayadurai & Eskandari, 2018). Because we now live in a very Mohamed M. Khalifa Tailab, Department of Finance and Investments, interconnected economy, this failure could have been trans- Lincoln University, 401 15th Street, Oakland, CA 94612, USA. mitted to other sectors. Therefore, evaluating the financial Email: mtailab@lincolnuca.edu Creative Commons CC BY: This article is distributed under the terms of the Creative Commons Attribution 4.0 License (https://creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/open-access-at-sage). 2 SAGE Open (Beltratti & Stulz, 2012). To explore this phenomenon, sev- predictors of business failure (Maricica & Georgeta, 2012), 15 eral papers such as Beltratti and Stulz (2012), Serrano-Cinca ratios for each bank were selected to build the initial model. et al. (2014), Adebambo et al. (2015), Cox et al. (2017), and According to the rule of thumb of model evaluation using PLS- Avkiran et al. (2018) investigated the impact of financial cri- SEM (Hair et al., 2016), only six ratios were employed each sis on banks’ performance from a variety of perspectives. year. The model assembles 15 indicators with four predecessor Thus far, however, insufficient attention has been paid by constructs (i.e., profitability of 2006, profitability of 2007, risk these studies to the role of prioritization of managerial deci- of 2006, and risk of 2007) and one final target construct (i.e., sions and policy activities in precipitating the financial crisis. profitability of 2008). Profitability and risk of 2007 mediate the Prioritization at a strategic and operational level is usually path of profitability and risk of 2006 and profitability of 2008. the difference between success and failure. This study helps The importance and performance values of profitability of address this gap in the literature by applying a technique to 2008’s predecessor constructs (i.e., profitability of 2006, profit- help bankers determine priority factors that should receive ability of 2007, risk of 2006, and risk of 2007) create the impor- their attention to prevent a repeat of the failure. tance-performance map of profitability of 2008. In light of this consideration, importance-performance The results plotted in the IPMA indicated that failed banks analysis, which is often called importance-performance map were predisposed to decreasing financial performance in analysis (IPMA) (Ringle & Sarstedt, 2016), has been found 2008 because of their poor performance in 2006 and 2007. to be a useful technique to identify the areas of improvement On the contrary, the profitability of 2006 and 2007 for non- that should be addressed by management activities (Martilla failed banks positioned them for increasing financial perfor- & James, 1977). This approach allows managers to improve mance in 2008. Although the IPMA can be conducted on the their management strategies because it indicates the main indicator level, this article limits its analysis by focusing on factors that require an immediate response (improvement) the construct level only. Finally, this research, to date, is the (Wyród-Wróbel & Biesok, 2017, p. 123). first to apply the IMPA using the secondary data collected The decision to apply this technique was driven by three from financial statements. motivations in particular: (a) IPMA facilitates more rigorous The rest of this article begins by presenting previous management decision-making; (b) IPMA is a powerful tool related work. This is followed by the research method where that can assist managers to set better priorities and better the procedures for predictive model assessment are outlined allocate scarce resources; and (c) having guidelines for per- in clear steps. Findings and discussion follow along with a formance assessment is as valuable to a firm as it is to the brief conclusion. individuals who invest in it (particularly during and follow- ing a financial crisis; Streukens et al., 2018). Literature Review To be clear, the purpose of this study is not to analyze the effect of financial crises on banks’ performance. Rather, it The use of PLS-SEM has received significant attention in introduces the IPMA and how to use it to provide a better several business disciplines (J. R. Barth et al., 2018). This understanding of where management should focus their technique is ubiquitous in marketing and management infor- attention. Reviewing business literature indicates that a mation systems, but it is still in its embryonic state in bank- plethora of papers have used IPMA in banking, such as ing literature because the advantages of this approach have Joseph et al. (2005), Ramayah et al. (2014), and Samar et al. yet to be discovered by the banking discipline (Avkiran, (2017). These papers utilized an end-user satisfaction survey 2018, p. 1). Only a few papers have used PLS-SEM in bank- to measure customers’ perceptions. This article is the first to ing literature. apply IPMA using secondary data collected from financial To my best knowledge, the first application of partial least statements to assess the performance of banks. IPMA was squares discriminant analysis (PLS-DA) for the prediction of applied to a unique data set of 140 failed U.S. banks that the 2008 U.S. banking crisis was done by Serrano-Cinca and closed in 2009 compared with the same number of nonfailed Gutiérrez-Nieto (2013). They compared PLS-DA with eight banks from 2006 to 2008. The U.S. banking data were used algorithms commonly used in bankruptcy prediction. It was as a case study because the crisis started in the United States, indicated that PLS-DA results resemble linear discriminant where many large banks lost most of their equity (Beltratti & analysis and support vector machine results. A particular Stulz, 2012). Thus, as U.S. banks play a major role in today’s advantage of this technique (PLS-DA) is that it is not affected global economy, their performance contributed more heavily by multicollinearity because it has been designed to deal to the crisis than other financial firms around the world. with this issue. They concluded that the interpretability of the It is well known that IPMA is the extension of partial least PLS-DA model was satisfactory. squares structural equation modeling (PLS-SEM; Ringle & The study by Serrano-Cinca et al. (2014) applied a path Sarstedt, 2016). The first requirement of the empirical work is model based on structural equations and logistic regression to to develop the PLS path model. The factors chosen for this investigate the financial symptoms that precede bankruptcy. model are those most relevant to success and failure: profit and They used low profitability, insufficient revenue, or low sol- risk. Because financial ratios are generally utilized as good vency ratios as proxies for symptoms, whereas loan growth Tailab 3 (some of them risky), specialization (real estate concentra- high risk might cause failure, whereas earning consecutive tion), and the pursuit of a turnover-driven strategy neglecting low profit puts management under pressure from sharehold- margin were used as proxies for the causes of these symp- ers. Thus, giving loans to customers is usually associated toms. They found that 5 years before the crisis, distressed with riskier loan practices, which can be measured by risk banks, compared with solvent banks, had the following: ratios. If the customers fail to repay those loans, profitability higher loan growth, higher concentration on real estate loans, will be affected. The initial model of this article is based on higher risk ratios, and higher turnover, but lower margins. the following assumption: If the years preceding the crisis Also, failed banks had a significant relationship between the were highly profitable, a bank would not fail because that percentage of real estate loans and risk. This relationship was profitability could be used to absorb future losses; in other negative in successful banks. Their findings confirmed that words, it could handle the risk. successful banks allocated real estate loans that were both The research model, which is depicted in Figure 1, com- fewer in number and higher in quality. I see this study as a bines five latent variables—profitability of 2006 (Prof ), good example of a causal analysis (i.e., symptoms and risk of 2006 (Risk ), profitability of 2007 (Prof ), risk of 2006 2007 causes), but the measurement invariance of the composite 2007 (Risk ), and profitability of 2008 (Prof )—as 2007 2008 model (MICOM)—which is a critical step in cross-group reflective constructs. For each reflective construct variable, investigation—has never been observed in this study. Without three manifests (indicators) were assigned. Profitability and measurement invariance, a study is susceptible to potential risk of 2007 mediate the path of profitability and risk of 2006 misspecification bias. Therefore, this article differs from the and profitability of 2008. Serrano-Cinca study in that the measurement invariance is Figure 1 presents both direct and indirect (mediating) tested before running a multiple analysis to compare the path effects that can be responsible for a bank’s distress. The prof- coefficients between failed and nonfailed banks. itability and risk in a previous year would affect the profit- To explain the drivers of bank soundness in G7 countries ability of the following year, and so on. The predictive model from 2003 to 2013, Ayadurai and Eskandari (2018) devel- covered the 3 years prior to the banks’ closing in 2009. oped a model with 17 indicators of six constructs as the As mentioned in the introduction, the IPMA focuses direct cause and eight as the indirect cause. They found that mainly on the key target constructs of interest in the PLS banks placed high importance on off-balance sheets and cap- path model. Figure 1 shows that the profitability of 2008 is ital activities, thus taking on higher risk. the final target construct, whereas the predecessor constructs By using PLS-SEM models, Avkiran et al. (2018) moni- are profitability of 2006, profitability of 2007, risk of 2006, tored the transmission of systematic risk from shadow banks and risk of 2007. Starting at the back end of Figure 1, it can to regular banks. The results of the predictive model indi- be seen that banks’ profitability in 2008 was positively or cated that a substantial degree of the variation in systematic negatively influenced by the predecessor constructs. By risk in the regulated banks was explained by micro-level and employing the PLS-SEM using SmartPLS 3 software (Hair macro-level linkages that can be traced to shadow banking. et al., 2016), the PLS model results can be utilized to calcu- In terms of obtaining forecasts for net charge-off rates for late the important scores. The important-performance values banks, J. R. Barth et al. (2018) used a PLS model to extract of the final target construct (i.e., profitability of 2008) were target-specific factors. They included more than 200 predic- created based on the important-performance values of the tor variables utilizing 250 quarterly macroeconomic data predecessor constructs (i.e., profitability of 2006, profitabil- collected from the period 1987: Q1 to 2016: Q4. The empiri- ity of 2007, risk of 2006, and risk of 2007). cal results showed that PLS outperformed benchmark mod- Figure 1 presents several paths (i.e., H1, H2, H3, H4, H5, els. They concluded that their model approach would assist H6, H7, and other indirect paths). In keeping with general banks in determining which variables cause failures and con- IPMA procedures, these path coefficients must be deter- tain losses to manageable levels. mined to be significant at any level of confidence before con- Although these papers presented the advantages of apply- firming that the required conditions for carrying out the ing PLS compared with traditional statistical methods, they IPMA have been established. did not apply the IPMA in this context. The present work, therefore, supplements the literature by reviewing the appli- Data Collection and Analysis Process cation of IPMA as an extension of PLS-SEM and testing the According to the FDIC, 140 American banks failed in 2009 MICOM. because of the financial crisis. To explore the reasons behind this failure and to contribute to the existing body of knowl- Research Method edge, this work analyzes the entire population of failed banks in 2009. This article compares those failed banks (service Path Model Estimation type = 0; n =140) with the nondistressed banks (service type The main goal of banks is to serve customers and obtain = 1; n =140) listed by the FDIC by employing both paramet- profit at the same time. To achieve this objective, banks must ric and nonparametric tests from 2006 to 2008. The total balance risk and return (J. R. Barth et al., 2018). Handling number of observations is 12,600 (280 firms × 3 years × 15 4 SAGE Open Figure 1. Direct and indirect (mediating) effects. indicators). According to Hair et al. (2016), the rule of thumb The measurement model of this study has five latent con- requires the sample size must be at least 10 times the maxi- structs, as depicted in Figure 1. Results of reliability analy- mum number of indicators associated with an outer model, ses using composite values show that all of these constructs which is met in this study. (profitability and risk) obtained a high reliability value above The financial data of this study were collected from the 0.7 thresholds (China et al., 2012). Table 1 shows that all FDIC and are available to the public. For each bank, 15 finan- indicators’ loadings are greater than 0.7 for both groups cial ratios were selected to build the model. Procedures were except for three indictors (Nimy for failed banks, and followed to evaluate the predictive model using PLS-SEM in Nimy and Nimy for nonfailed banks). As evidenced, 2007 2008 SmartPLS 3 software (Hair et al., 2016). Accordingly, the the low loading value of those three indicators ranges from financial ratios were filtered by eliminating those with a weak 0.5 to ≈ 0.7. In exploratory studies, loadings higher than 0.4 effect and using only the ratios with a strong effect to predict are acceptable (Avkiran et al., 2015; Hair et al., 2013, p. 6). failure. According to the rules of thumb of model evaluation This confirms that the internal consistency requirements using PLS-SEM, only six ratios, as defined in Table A1, were were established. employed for each year. Net interest margin (Nimy), return To validate the measurement model, two types of validity on assets (ROA), and return on equity (ROE) are proxy for were checked: convergent validity and discriminant validity. profitability, whereas core capital (leverage) ratio (Rbc1aaj), The average variance extracted (AVE) was used as a crite- Tier 1 risk-based capital ratio (Rbc1rwaj), and total risk-based rion of convergent validity, whereas the Fornell–Larcker cri- capital ratio of risk (Rbcrwaj) are proxy for the capital terion and cross-loadings were utilized as criteria of strength of the bank. To evaluate the univariate discrimina- discriminant validity. tory power of each ratio, a mean test was computed. In establishing convergent validity, a threshold of accept- ability of greater than 0.5 was ensured for the value of AVE. (China et al., 2012; Hair et al., 2016; Rasoolimanesh et al., Model Assessment Utilizing PLS-SEM 2017). Findings presented in Column 4 of Table 1 indicate Before reporting the findings of this work, reliability and that all AVE values met the requirements for the construct validity must be assessed as a prerequisite for accurate esti- measures’ convergent validity, which is above the accept- mation in SEM (Nitzl, 2016). Although previous works offer ability value. The convergent validity ranges from 0.625 to criteria to assess partial model structures, this article follows 0.969. The results showed that the loading’s value of the guideline of Hair et al. (2016). To prevent a common mis- Nimy , Nimy , and Nimy ranged from 0.4 to 0.7, 2006 2007 2008 take with model assessment, the reliable and valid measure- validating the decision to retain those indicators in the model. ment of reflective indicators should not be applied to Finally, the article measures discriminant validity by formative indicators (L. Lee et al., 2011). looking at the square root of AVE for each latent variable by Tailab 5 Table 1. Evaluation Results of the Measurement Model. 2 3 4 Loading Composite reliability AVE 1 Failed Nonfailed Failed Nonfailed Failed Nonfailed Prof 0.880 0.908 0.713 0.768 Nimy 0.673 0.792 ROA 0.882 0.908 ROE 0.953 0.923 Risk 0.977 0.971 0.934 0.917 Rbc1aaj 0.931 0.910 Rbc1rwaj 0.989 0.984 Rbcrwaj 0.978 0.976 Prof 0.901 0.851 0.754 0.669 Nimy 0.777 0.523 ROA 0.906 0.936 ROE 0.914 0.926 Risk 0.988 0.965 0.965 0.901 Rbc1aaj 0.969 0.896 Rbc1rwaj 0.989 0.978 Rbcrwaj 0.989 0.971 Prof 0.914 0.827 0.780 0.625 Nimy 0.842 0.531 ROA 0.924 0.891 ROE 0.881 0.894 Note. AVE = average variance extracted; ROA = return on assets; ROE = return on equity. Table 2. Discriminate Validity (Fornell–Larcker Criterion). Failed banks Nonfailed banks Constructs Prof Prof Prof Risk Risk Prof Prof Prof Risk Risk 2006 2007 2008 2006 2007 2006 2007 2008 2006 2007 Prof .844 .876 Prof −.239 .868 .875 .818 Prof −.297 .337 .883 .563 .725 .790 Risk .103 −.174 −.254 .966 −.575 −.430 −.161 .957 Risk −.140 .414 .189 .018 .982 −.451 −.331 −.119 .913 .949 Note. The values in bold are the square root of AVE, and below the diagonal are the construct correlations. AVE = average variance extracted. applying Fornell–Larcker criterion. Table 2 shows that the discriminant validity is established (Henseler et al., 2015, square root of AVE for each latent variable (in bold) is higher 2016); however, if the value of the HTMT is greater than this than the absolute values of other correlation values among threshold, there is a lack of discriminant validity (Ngah et al., the latent variables. This indicates that all of the discriminant 2018). Table 3 shows that all HTMT values were lower than validity requirements through the square root of AVE to the the threshold value. correlation, and through the cross-loading analysis, are met. In addition to the validity evaluation, this work checks for Although the Fornell–Larcker criterion and cross- multicollinearity by variance inflation factor (VIF) and predic- loadings are widely used for evaluating discriminant validity, tion relevance (Q ). A VIF of greater than 5 suggests a multi- SmartPLS provides yet another reliable criterion called the collinearity issue, and a VIF of greater than 10 reveals a critical heterotrait-monotrait ratio (HTMT). In using the HTMT cri- multicollinearity problem. For stricter tests, VIF should be terion to evaluate discriminant validity, one must determine below 3.3 in the inner model between latent variable predictors the HTMT value for all pairs of reflective constructs. If the (Cenfetelli & Geneviève, 2009, p. 694; Kock & Lynn, 2012; value of HTMT is below .85 (the threshold value), then Lowry & Gaskin, 2014, p. 137). Table 4 presents the outcome 6 SAGE Open Table 3. Discriminate Validity (HTMT Criterion). Constructs Prof Prof Prof Risk Risk 2006 2007 2008 2006 2007 Prof Prof .275 Prof .102 .574 Risk .122 .094 .105 Risk .139 .294 .179 .637 Note. All findings satisfy the HTMT.85 criterion. HTMT = heterotrait-monotrait ratio. 2 2 Table 4. Measurement of Interaction Through Effect Size f and Q . Prof Risk Prof Q² 2007 2007 2008 2 2 2 VIF f VIF f VIF Total sample Failed Nonfailed Constructs f Prof .056 1.013 .004 1.105 Prof .191 1.268 .035 .045 .489 Risk .002 1.013 .636 1.000 .016 1.801 Risk .014 1.966 .349 −.001 .724 Prof .159 .138 .305 SRMR .08 Note. The results of profitability and risk of year (t = 2006) as exogenous latent variables on profitability and risk of year (t + 1 = 2007) and (t + 2 = 2008) as endogenous variables, and the results of profitability and risk of year (t + 1 = 2007) as exogenous variables on profitability of year (t + 2 = 2008) as endogenous variables. VIF = variance inflation factor; SRMR = standardized root mean square residual. of a multicollinearity test for a structural model (inner model) To measure the magnitude of R values as a criterion for that indicates collinearity is not a concern. However, five predictive relevance, this article tests Stone and Geisser’s Q reflective indicators’ values (outer model) exceed the thresh- to be sure that the model has predictive relevance. The Q old of 5. So, multicollinearity arising among these indicators was calculated based on the blindfolding procedure. A posi- has a small correlation with the construct latent variables. tive Q value gives evidence that the path model has predic- Collinearity among these indicators occurs due to the type of tive relevance, whereas a negative value signifies a lack of financial data. In particular, financial ratios that share a com- predictive relevance (Henseler et al., 2009, p. 303). The mon numerator or denominator can suffer from multicol- result demonstrated in Table 5 indicates that all endogenous linearity (Serrano-Cinca & Gutiérrez-Nieto, 2013). We constructs were above zero (positive), which confirms that expected that some degree of collinearity would occur because the path model has predictive relevance. By contrast, Risk the corresponding numerators and denominators are likely to for failed banks has no predictive relevance as Q stands at a be correlated. negative value. R was used to evaluate predictive power. Results indicate that the value of R was considered weak for the whole sam- Measurement Invariance and Multi-Group 2 2 ple (R = .23) and for failed banks (R = .20), whereas it was Analysis considered moderate for nonfailed banks (R = .56) (Chin, 1998). In this endeavor, however, Henseler et al. (2009) Measurement invariance, which is also known as measure- accept moderate R if the model has only one or two exoge- ment equivalence, is a crucial step in cross-group investiga- nous latent variables. It is very difficult to find accepted rules tion (Ruzzier et al., 2014). This technique enables researchers for preferred R because it is usually dependent on the degree to identify whether parameters of the structural model and of model complexity and research discipline (Hair et al., measurement model are equivalent (i.e., invariance) across 2016, p. 199). two or more groups (China et al., 2012). Measurement invari- Unlike R , which looks at each endogenous latent vari- ance is needed to be sure that the differences are not attribut- able, f measures whether the effect of a particular exogenous able to measurement model differences across the groups variable on an endogenous variable is substantive when that (Kock, 2017). If one failed to establish invariance, it would be exogenous variable is omitted from the model. This test is difficult to determine whether the differences observed were called the effect size. According to the Hair et al. (2016, due to true differences (China et al., 2012, p. 1). p. 201) guideline, f values of .02, .15, and .35, respectively, Before running an multiple-group analysis (MGA) to represent small, medium, and large. compare the path coefficients between failed and nonfailed Tailab 7 Table 5. Results of Invariance Measurement Testing Using Permutation. Compositional invariance Configural Partial Equal mean value Equal variance Full invariance (same measurement measurement algorithms for 5% quantile invariance invariance Constructs both groups) C = 1 of Cu p Values established Difference CIs Difference CIs established Prof Yes .996 [.921, 1.000] .684 Yes −0.544 [−.202, .202] 0.157** [−0.331, 0.325] No/Yes Prof Yes .998 [.983, 1.000] .525 Yes −0.682 [−0.204, 0.209] 1.187 [−0.288, 0.295] No/No Prof Yes .999 [.989, 1.000] .500 Yes −0.463 [−0.207, 0.187] 1.506 [−0.313, 0.326] No/No Risk Yes 1.000 [.999, 1.000] .780 Yes −0.795 [−0.201, 0.202] −0.418 [−0.337, 0.321] No/No Risk Yes 1.000 [1.000, 1.000] .300 Yes −0.856 [−0.205, 0.203] −0.305* [−0.330, 0.334] No/Yes Note. If C exceeds the 5% quantile of Cu, compositional invariance is established CI = confidence interval. *p value greater than .05 indicates that the correlation is not significantly lower than 1. **There are no significant differences in the mean and variance values of the latent variables across the two groups because the original difference value is within the corresponding CI. banks, it is important to employ invariance testing to avoid extensive work of Ringle, Sarstedt, and Mooi (2010) and potential misspecification bias and misleading results. The Ringle, Wende, and Will (2010), and implemented in the MICOM test is required in PLS-SEM to be confident that the software application SmartPLS (Sarstedt et al., 2011, p. 35). group differences in the model “do not result from either dis- This approach combines a finite mixture procedure with tinctive content or meanings of the latent variables across an expectation- groups or from the measurement scale” (Sarstedt et al., 2018, maximization (EM) algorithm (Loureiro & Miranda, 2011). p. 209). The MICOM analysis consists of three steps SmartPLS 3 (Hair et al., 2018) was used to segment the sam- (Henseler et al., 2016, p. 412): (a) configural invariance, ple based on the estimated scores for latent variables. which is the first and weakest level of measurement (Kim According to the result of FIMIX-PLS presented in Table 6, et al., 2013, p. 502); (b) compositional invariance; and (c) the the analysis considers two-, three-, and five-segment solu- equality of composite mean values and variances. Thus, if tions. Applying the relevant assessment criteria provided by configural invariance and compositional invariance are Hair et al. (2018), two segments was the appropriate choice. established, partial measurement invariance is confirmed. Because the optimal solution is the number of segments with After that, researchers can compare the path coefficient esti- the lowest value (bold numbers) and the highest value of mates between groups. In addition, if partial measurement entropy values (EN) (Hair et al., 2018, p. 196), Segments 1 invariance is established and the composites have equal and 4 have been eliminated. No single criterion indicates the mean values and variances (Step 3) “across the groups, full same number of segments, whereas the EN range between measurement invariance is confirmed, which supports the 0.683 and 0.831, which are higher than the threshold of 0.5 pooled data analysis” (Henseler et al., 2016, p. 413). Pooled (Hair et al., 2018, p. 197; Schlagel & Sarstedt, 2016, p. 640). data analysis examines whether a common core of relation- Panel B of Table 6 indicates that Segment 5 has very small ships exists across the groups (Durvasula et al., 1993). segment sizes. To warrant valid analysis, Segment 5 was In accordance with the MICOM analysis, Table 5 shows dropped. Panel C of Table 6 confirms that the R value in that partial measurement invariance is established, which is a Segment 2 is the highest compared with the original sample. requirement to compare groups and test any significant It can be concluded that heterogeneity in this study is not differences. prevalent. Robustness Check IPMA To ensure that validity measurement is established when IPMA—also called importance-performance matrix, impact- using PLS-SEM, checking unobserved heterogeneity is very performance map, and priority map analysis (Ringle & important (Hair et al., 2018; Sarstedt et al., 2018; Schlagel & Sarstedt, 2016, p. 1866)—was first proposed and introduced Sarstedt, 2016). Although data of this study are obtained by Martilla and James (1977). The IPMA approach examines from the same population (the banking industry), assumption not only the performance of an item but also the importance of homogeneity is unrealistic because individuals are not of that item. The objective of this analysis is to identify the homogeneous in their perceptions and evaluations of unob- (unstandardized) total effect of predecessor construct’s served constructs (Ansari et al., 2000). This article applies importance (e.g., profitability of 2006) in anticipating a spe- the finite mixture partial least squares (FIMIX-PLS) method, cific target endogenous construct (e.g., profitability 2008) which was proposed by Hahn et al. (2002), advanced by the (Hair et al., 2016, p. 276; Hair et al., 2018, p. 105). The total 8 SAGE Open Table 6. Model Selection. Fit indices Number of segments Criteria 2 3 5 Full data set Panel A: Fit indices for segment solution AIC (Akaike’s information criterion) 1,770.361 1,726.119 1,671.359 AIC3 (modified AIC with Factor 3) 1,791.361 1,758.119 1,725.359 AIC4 (modified AIC with Factor 4) 1,812.361 1,790.119 1,779.359 BIC (Bayesian information criteria) 1,846.691 1,842.432 1,867.638 CAIC (consistent AIC) 1,867.691 1,874.432 1,921.638 HQ (Hannan–Quinn criterion) 1,800.977 1,772.772 1,750.087 MDL5 (minimum description length with Factor 5) 2,320.014 2,563.685 3,084.752 EN (entropy statistic [Normed]) 0.831 0.683 0.733 Panel B: Relative segment sizes Segment sizes, % 0.292* 0.182 0.083 Panel C: FIMIX-PLS, R Prof .924 .160 .810 .058 Prof .924 .504 .975 .233 Note. FIMIX-PLS = finite mixture partial least squares. Figure 2. Importance-performance matrix and path model with results. effect demonstrates the importance of apparent variables, performance (Jaafar et al., 2016; Y.-C. Lee et al., 2008). The whereas the mean value of their scores (ranging from 0, goal is to determine each predecessor construct’s importance which is considered the lowest, to 100, the highest) reflects in terms of its total impact on each target endogenous con- their performance (Höck et al., 2010, p. 201). The interpreta- struct (performance). tion of IPMA is that a 1-unit increase in the predecessor’s Figure 2 shows that the final target construct—profitability performance (e.g., Prof ) increases the performance of the of 2008—was affected both directly and indirectly by the target construct (Prof ) by the size of the predecessor’s profitability and risk of 2006 and 2007. unstandardized total effect (Hair at el., 2016, p. 278). The Starting at the back end of Figure 2, when we look at the IPMA technique has two dimensions: importance and final target construct (i.e., profitability of 2008), profitability Tailab 9 Figure 3. Importance performance map of the profitability of 2008 (failed banks). of 2007 has a relatively high positive importance with the significantly different across the two groups (failed and non- path coefficient of 0.456, whereas profitability of 2006 has failed banks) except for one indicator: Nimy. lower importance with the path coefficient of 0.056. The Descriptive analysis depicted in Table 7 confirms that the path coefficient, depicted as arrows, demonstrates the rela- average profitability was higher for nonfailed than failed tive importance, whereas the performance values, depicted banks. Unlike distressed banks, nonfailed banks had never as circles, are the average values of latent variables’ scores had a negative performance, which protected them from on a scale of 1 to 100. It should be noticed that a score closer bankruptcy. Failed banks had a lower average risk than non- to 100 indicates a higher performance latent variable (Hair failed banks. Hence, low profitability can be a symptom of et al., 2016). failure. The results of IPMA presented in Figure 2 are also depicted in Figures 3 (failed banks) and 4 (nonfailed banks) Test of Significant Paths and discussed more deeply later. A comprehensive under- standing of how to read and use results plotted in those fig- As with any other analytical technique, IPMA needs to be ures can assist management in improving low performance driven by statistical power consideration (Streukens et al., by focusing on high importance (Hock et al., 2010). 2018, p. 380). Primary data analysis confirms interesting The IPMA approach must meet two requirements before findings in six significant paths related to the main objective any application: (a) all indicators must have the same orien- of this article. The results are summarized in Table 8 for each tation, and (b) outer weight must not be negative (Hair et al., subgroup. It is important to consider the sign of the path coef- 2016, p. 208; Hair et al., 2018, p. 123; Ringle & Sarstedt, ficient (positive or negative), as it can be the reason those 2016, p. 1868). These requirements have been met, as shown banks went bankrupt (e.g., negative profit). The path coeffi- in Tables B2 and B3. cients were tested (using 5,000 bootstrap) by applying a one- tailed test (for a more detailed description of when or whether to utilize one-tailed or two-tailed tests, see Kock, 2015). Results The study found that the profitability in a given year was significantly correlated to the profitability of a following Descriptive Statistic year. In failed banks, it was found that the profitability of Table 7 presents the descriptive statistics of our full sample, 2006 (Prof → Prof , β = −0.233, p < .000) was nega- 2006 2007 as well as of subsamples of financial performance. Results tively associated with the profitability of 2007. In other reported in Table 7 confirm that all indicators (ratios) are words, for nonfailed banks, the study observes a strong 10 SAGE Open Table 7. Summary Statistics of the Two Groups (Failed and Nonfailed) and the Whole Sample. Total sample (N = 280) Failed banks (N= 140) Nonfailed banks (N = 140) Univariate test Construct p-value mean variables Indicators M SD M SD M SD test Profitability Nimy 0.038 0.011 0.037 0.01 0.04 0.01 .238 ROA −0.001 0.024 −0.009 0.025 0.01 0.02 (.000) ROE 0.048 0.11 0.003 0.116 0.09 0.08 (.000) Risk Rbc1aaj 0.101 0.043 0.089 0.039 0.11 0.04 (.000) Rbc1rwaj 0.13 0.054 0.107 0.042 0.15 0.05 (.000) Rbcrwaj 0.139 0.048 0.119 0.039 0.16 0.05 (.000) Profitability Nimy 0.038 0.011 0.037 0.012 0.04 0.01 .222 ROA −0.003 0.027 −0.014 0.033 0.01 0.01 (.000) ROE 0.04 0.107 −0.001 0.117 0.08 0.08 (.000) Risk Rbc1aaj 0.096 0.039 0.082 0.036 0.11 0.04 (.000) Rbc1rwaj 0.125 0.052 0.101 0.041 0.15 0.05 (.000) Rbcrwaj 0.134 0.048 0.113 0.038 0.16 0.05 (.000) Profitability Nimy 0.037 0.011 0.037 0.014 0.04 0.01 .366 ROA −0.005 0.029 −0.014 0.037 0 0.01 (.000) ROE 0.028 0.107 0.004 0.125 0.05 0.08 (.000) Risk Rbc1aaj 0.087 0.039 0.073 0.042 0.1 0.03 (.000) Rbc1rwaj 0.114 0.053 0.089 0.045 0.14 0.05 (.000) Rbcrwaj 0.124 0.051 0.1 0.045 0.15 0.04 (.000) Note. Path coefficient is significant at p < .05 for one-tailed test. ROA = return on assets; ROE = return on equity. Table 8. Results of Hypothesis Testing. Path coefficient CIs bias-corrected Failed Nonfailed Failed Nonfailed Hypotheses Relationship β t value β t value 5.0% 0.95% 5.0% 0.95% Supported HI Prof → Prof −0.233* 2.937 0.938* 20.807 [−0.338, −0.076] [0.858, 1.003] Yes/Yes 2006 2007 H2 Prof → Prof −0.214* 3.099 −0.174 0.893 [−0.323, −0.093] [−0.507, 0.130] Yes/No 2006 2008 H3 Prof → Prof 0.223* 2.662 0.947* 5.987 [0.072, 0.353] [0.690, 1.213] Yes/Yes 2007 2008 H4 Risk → Prof −0.151* 1.683 0.109 1.629 [−0.286, −0.001] [−0.012, 0.208] Yes/No 2006 2007 H5 Risk → Prof −0.194* 2.577 0.243 0.927 [−0.319, −0.071] [−0.134, 0.719] Yes/No 2006 2008 H6 Risk → Risk 0.018 0.178 0.913* 31.860 [−0.125, 0.202] [0.858, 0.955] No/Yes 2006 2007 H7 Risk → Prof 0.070 0.702 −0.106 0.447 [−0.092, 0.237] [−0.528, 0.257] No/No 2007 2008 H8 Prof → Profi → Prof −0.054* 1.797 0.895* 5.321 [−0.100, −0.011] [0.636, 1.187] Yes/Yes 2006 2007 2008 H9 Risk → Prof → Prof −0.034 1.405 0.097 1.561 [−0.082, −0.003] [−0.012, 0.208] No/No 2006 2007 2008 H10 Risk → Risk → Prof 0.001 0.103 −0.097 0.443 [−0.010, 0.035] [−0.490, 0.236] No/No 2006 2007 2008 Note. CI = confidence interval. *p < .05 for one-tailed test because the sign “+” or “−” is important. positive association between the profitability of 2006 the profitability of 2007 for failed banks decreased by (Prof → Prof , β = 0.938, p < .000) and the profit- −0.233, whereas the profitability of nonfailed banks 2006 2007 ability of 2007. The regression coefficient indicates that increased by 0.938. when the profitability of 2006 changed by 1 unit (dollar), Furthermore, in both groups, there was a positive and sig- with the assumption that other factors remained constant, nificant effect of the profitability of 2007 on the profitability Tailab 11 Table 9. Results of Hypothesis Testing (MGA-PLS). Path coefficient Welch– Hypothesis Path differences p-value MGA Parametric test Satterthwaite test Supported H1 Prof → Prof 1.161 1.00** 0.000* 0.000* Yes 2006 2007 H2 Prof → Prof 0.040 0.59 0.847 0.847 No 2006 2008 H3 Prof → Prof 0.723 1.00** 0.000* 0.000* Yes 2007 2008 H4 Risk → Prof 0.261 0.99** 0.018* 0.019* Yes 2006 2007 H5 Risk → Prof 0.437 0.95 0.107 0.108 No 2006 2008 H6 Risk → Risk 0.895 1.00** 0.000* 0.000* Yes 2006 2007 H7 Risk → Prof 0.176 0.26 0.495 0.495 No 2007 2008 H8 Prof → Prof → Prof 0.938 1.00** 0.000* 0.000* Yes 2006 2007 2008 H9 Risk → Prof → Prof 0.039 0.584 0.865 0.865 No 2006 2007 2008 Note. **p value of MGA lower than .05 or higher than .95 indicates significant differences at the 5% level. Parametric and Welch–Satterthwaite tests are significant for. *p value only at the 5% level, which is lower than .05. PLS-SEM = partial least squares structural equation modeling; MGA = multiple- group analysis. This indicates the difference between the group-specific path coefficients, which is not necessary to be lower than 1 (Christian Ringle, Personal communication, April 17, 2018). of 2008 (Prof → Prof , β = 0.223, p < .000 and β = To measure the difference between the two groups (failed 2007 2008 0.947, p < .000) for failed and nonfailed banks, respectively. and nonfailed banks), the study employed Henseler’s MGA, There was significant evidence that the profitability of 2006 the parametric test, and the Welch–Satterthwaite test. had a negative impact on the performance of 2008 for failed Findings illustrated in Table 9 reveal significant differences banks (Prof → Prof , β = −0.214, p < .000). This was between failed and nonfailed banks in five paths. First, the 2006 2008 not the case for nonfailed banks because the study could not significant difference for Prof → Prof (H1) was 1.161. 2006 2007 find a significant correlation. This pointed to the difference between the group-specific In terms of risk effect, the current study indicated several path coefficients, which need not be lower than 1.0 to estab- significant paths. First, the path coefficient for Prof → lish. For example, if the path coefficient is 0.5 for one group Risk was significantly negative (β = −0.151, p < .000) and −0.6 for another group, then the difference between the for failed banks. The path coefficient for Prof → Risk two groups is 1.1 (Christian Ringle, personal communica- 2008 2006 was also significantly negative (β = −0.194, p < .000). tion, April 17, 2018). Second, the path coefficient for Prof These paths were not significant for nonfailed banks. This → Prof (H3) differed between the two groups by 0.723, result suggested that for failed banks when the risk changed which was significant according to the p value of Henseler’s by 1 unit, it tended to decrease the profitability. This was not MGA and permutation and Welch–Satterthwaite tests. Third, the case for nonfailed banks. Therefore, it can be assumed the path coefficient for Prof → Risk (H4) supported a 2007 2006 that this was the potential reason behind failure. However, significant difference of 0.261. In addition, the path coeffi- the path coefficient for Risk → Risk was significantly cient for Risk → Risk (H6) revealed a significant dif- 2006 2007 2007 2008 positive (β = 0.913, p < .000) for nonfailed banks (only). ference between the two groups of 0.895. Finally, the results Finally, the study revealed no significant coefficient for indicated that one of the two mediating paths was signifi- Prof → Risk for both groups. Thus, the effect of cantly different. The indirect path Prof → Prof → 2008 2007 2006 2007 Risk on Prof was unsupported in both groups. Prof (H8) differed by 0.938. The results did not support a 2007 2008 2008 The indirect (mediating) effect’s findings illustrated that significant difference with regard to the effects of Prof → one of the three mediating paths was significant. Findings Prof (H2), Risk → Prof (H5), or Risk → Prof 2008 2006 2008 2006 2007 illustrated in Table 8 show that the indirect path for Prof → Prof (H9). 2006 2008 → Prof → Prof was significantly negative for failed 2007 2008 banks (β = −0.054, p < .000), whereas it was highly posi- Results of IPMA. With bootstrapping with 5,000 subsamples tive for nonfailed banks (β = 0.895, p < .000). However, confirming that some of the path coefficients were statisti- other specific indirect effects were not significant. This cally significant and the requirements for carrying out the result indicated that consecutive low profitability of failed IPMA approach have been met, the findings of IPMA can banks caused the failure, unlike the consecutive high profit- finally be discussed. These results are plotted in Figure 3 ability of nonfailed banks that was used to absorb future (i.e., failed banks) and Figure 4 (i.e., nonfailed banks). In losses. This result reflected the difference in strategy each importance-performance map, the analysis concen- between failed and nonfailed banks. The study observed no trated on the lower right area to enhance improvement significant mediating role of Prof and Risk between because items plotted in that area have high importance with 2007 2007 Risk and Prof . low performance. Concentrating constructive action in this 2006 2008 12 SAGE Open Figure 4. Importance performance map of the profitability of 2008 (nonfailed banks). area will produce maximum results (Martilla & James, 1977, Although Figure 4 presents that the profitability of 2006 p. 78). For failed banks, the results confirmed that the profit- was the second-highest importance indicator, it was not a ability of 2007 and risk of 2007 had particularly high impact significant important variable in the projection of the perfor- on the profitability of 2008. mance of 2008. It is evident from Figure 3 that the profitability of 2007 had a large impact on the profitability of 2008 in failed banks, Discussion and Managerial Implications and thus represents a major opportunity for improvement that could have been addressed by bankers’ activities. More The findings suggest that there is a significant difference precisely, the importance of the profitability of 2007 had a between failed and nonfailed banks regarding the effect of positive total effect on 2008: a 1-dollar increase in the profit- the financial crisis in 2007 and 2008. Research on banks’ ability of 2007 increased the profitability of 2008 by 0.254, performance during the crisis (Beltratti & Stulz, 2012) whereas changing the risk by 1 dollar in 2007 tended to showed that banks were affected differentially because they increase the profitability of 2008 by 0.047. This indicated had different balance sheets and profitability before the cri- that there was indeed room to improve the performance of sis. For failed banks, negative profitability measured by 2008. As seen in Figure 3, the profitability of 2006 had the ROA and a fall in ROE appeared 3 years before failure. strongest negative total effect on the profitability of 2008. Bankers should have dealt with this trend as a symptom of This variable was very important to affect the performance failure. This result is consistent with Beltratti and Stulz of 2008 because the total effect was significant. This study (2012), who confirmed that bank profitability in 2006 played found that the total indirect effect of the profitability of 2006 a more significant role in determining bank performance dur- on the performance of 2007 through 2008 was negative ing the crisis than did other factors such as bank governance (–0.056). Both direct and indirect total effects confirmed that and bank regulation. This was not apparent in the case of banks failed in 2008 because of their financial performance nonfailed banks whose profitability was never negative. For in 2006. nonfailed banks, increasing the income tended to increase It is evident from Figure 4 that for nonfailed banks, the the return, which was reflected in profitability ratios as a highest performance indicator was the profitability of 2007. consequence of healthy growth. This result is in line with the This factor was confirmed to be a significant predictor of the accounting theory that assumes that firms can use their net profitability of 2008. Unlike the failed banks, IPMA showed income to absorb future losses. that a 1-dollar increase in the profitability of 2007 increased Turning to performance and risk-taking, it can be seen the profitability of 2008 by 0.887 cents. that the risk that was taken in 2006 led to the poor Tailab 13 performance of failed banks over the next 2 years. However, the performance of 2007 by using their net income to either the risk that was taken by nonfailed banks in 2006 had no recover their previous losses or absorb future losses. correlation with bank outcomes. This is a striking result—it In particular, this study underscores that the most detri- was observed that the average risk (e.g., core capital (lever- mental managerial decisions and policy activities by failed age) ratio, Tier 1 risk-based capital ratio, and total risk-based banks were in regard to high-risk loans. The study indicates capital ratio) accepted by failed banks was lower than the that the risk that was taken in 2007 should have received average risk accepted by nonfailed banks. That lower aver- significant attention because it had a positive total effect on age was the cause of U.S. bank failures during the financial the profitability of 2008 and represented an opportunity for crisis. This finding agreed with the study by Beltratti and failed banks to increase profitability in 2008. Unfortunately, Stulz (2012), who found that large banks with more Tier 1 bankers missed this remarkable opportunity to avert disaster capital and more deposit financing at the end of 2006 had before the crisis arrived. In consequence, loans went unpaid, significantly higher returns during the crisis. This evidence is profitability went down, and banks closed. Applying IPMA also consistent with Serrano-Cinca et al. (2014) who reported allows companies to determine the best allocation of their that nonfailed banks compensated for increases in risk by resources based on the results of the IPMA. The interpreta- strengthening their core capital. However, Cox et al. (2017) tion of the comparison results between failed and nonfailed found that banks failed during the 2008–2010 financial crisis banks was satisfactory because test measurement invariance because they accepted more risk, specifically by having was done to avoid potential misspecification bias. higher financial leverage. This is in line with Serrano-Cinca In closing, managers who use and apply IPMA will et al. (2014), who showed that 5 years before the crisis, failed obtain useful conceptual insights by overlaying the impor- banks had higher loan growths, higher concentration on real tance-performance analysis to prioritize their financial estate loans, higher risk ratios, and higher turnover, but lower decision-making. margins. It can be concluded that the risk that was taken a few years before the crisis had a significant effect on a bank’s Conclusion performance. It remains to be determined why risk taken in 2007 was not significantly associated with profitability in Overall, this research addresses an overview of the applica- 2008. tion of the IPMA as a useful technique to expand the analysis This article extends the application of PLS-SEM by using of PLS-SEM results. This work develops a model to help IPMA to determine priority factors and should be useful to researchers who are keen on applying IPMA in the banking managers seeking to improve banking performance (espe- and finance fields. The results of IPMA indicated that failed cially managers of those banks that failed). The predictive banks were predisposed to decreasing financial performance model developed for this study indicated that failed banks in 2008 because of poor performance in 2006. Conversely, were predisposed to decreasing financial performance in the profitability of 2006 and 2007 positioned nonfailed banks 2008 because of poor performance in 2006. Conversely, the for increasing financial performance in 2008. It must be profitability of 2006 and 2007 positioned nonfailed banks for borne in mind that this research was only conducted on a increasing financial performance in 2008. For failed banks, small group of failed and nonfailed U.S. banks over a short financial performance in 2006 was found to be the highest period. Further research is thus needed to investigate the per- negative important factor that bankers should have addressed formance of large banks around the world during the crisis to avoid failure. Failed banks should have taken advantage of before generalized conclusions can be drawn. 14 SAGE Open Appendix A Table A1. Definition of Variables. Portability Nimy Net interest margin Total interest income minus the total interest expense as a percentage of average earning assets ROA Return on assets Net income after taxes and extraordinary items as a percentage of average total assets. ROE Return on Equity Annualized net income as a percentage of average equity on a consolidated basis. Risk Rbc1aaj Core capital (leverage) ratio Tier 1 (core) capital as a percentage of average total assets minus ineligible intangibles Rbc1rwaj Tier 1 risk-based capital ratio Tier 1 (core) capital as a percentage of risk-weighted assets as defined by the appropriate federal regulator for prompt corrective action during that time period Rbcrwaj Total risk-based capital ratio Total risk-based capital as a percentage of risk-weighted assets Note. ROA = return on assets; ROE = return on equity. Appendix B Table B1. Unstandardized Outer Weights. Prof Risk Prof Risk Profit 2006 2006 2007 2007 2008 Indictor Failed Nonfailed Failed Nonfailed Failed Nonfailed Failed Nonfailed Failed Nonfailed Nimy 0.28 0.36 ROA 0.32 0.37 ROE 0.55 0.41 Rbc1aaj 0.32 0.33 Rbc1rwaj 0.37 0.36 Rbcrwaj 0.35 0.36 Nimy 0.34 0.28 ROA 0.43 0.47 ROE 0.38 0.45 Rbc1aaj 0.36 0.34 Rbc1rwaj 0.33 0.36 Rbcrwaj 0.33 0.35 Nimy 0.35 0.34 ROE 0.42 0.45 ROA 0.35 0.46 Note. ROA = return on assets; ROE = return on equity. Tailab 15 Table B2. Rescaled. Prof Risk Prof Risk Prof 2006 2006 2007 2007 2008 Failed Nonfailed Failed Nonfailed Failed Nonfailed Failed Nonfailed Failed Nonfailed Nimy 0.61 0.57 ROA 0.28 0.34 ROE 0.11 0.09 Rbc1aaj 0.31 0.34 Rbc1rwaj 0.34 0.31 Rbcrwaj 0.35 0.35 Nimy 0.64 0.41 ROA2007 0.28 0.51 ROE 0.07 0.08 Rbc1aaj 0.37 0.39 Rbc1rwaj 0.31 0.29 Rbcrwaj 0.32 0.32 Nimy 0.64 0.51 ROA 0.29 0.41 ROE 0.07 0.08 Note. ROA = return on assets; ROE = return on equity. 6. In accordance with the basic PLS algorithm procedure, the Acknowledgments results presented in this section were obtained after Iteration I would like to thank Mike Tsionas (editor), four anonymous refer- 13; the bootstrapping procedure with 5,000 resamples with ees, and Themistoclis Pantos for their insightful, supportive feed- one-tailed t test and the blindfolding procedure with an omis- back, and comments on this research paper. I also thank Ms. Ashwini sion distance of 7 to Prof as the final target constructs were Thakur from Lincoln University for her research assistance. applied. 7. These criteria are generally consistent with formative latent Declaration of Conflicting Interests variable theory (for more details, see Kock, 2014, p. 11). The author(s) declared no potential conflicts of interest with respect 8. Using the value of R to evaluate model fit might allow to the research, authorship, and/or publication of this article. researchers to describe a model with good fit as poor. Thus, it can be said that fitness measures should only be based on how well the parameters are able to match the sample covariance Funding (Chin, 2010, p. 657). The author(s) disclosed receipt of the following financial support 9. It should be noted that a low value represents a negative out- for the research, authorship, and/or publication of this article: This come, whereas a high value represents a positive outcome. work was supported by the President’s Faculty Grant Program, However, it cannot be concluded that higher latent variables Lincoln University. indicate better performance (Hair et al., 2016, p. 280). ORCID iD References Mohamed M. Khalifa Tailab https://orcid.org/0000-0001- Adebambo, B., Brockman, P., & Yan, X. (2015). Anticipating the 6521-7487 2007–2008 financial crisis: Who knew what and when did they know it? Journal of Financial and Quantitative Analysis, Notes 50(4), 647–669. 1. This is the main reason why this article adopts partial least Ansari, A., Jedidi, K., & Jagpal, S. (2000). A hierarchical Bayesian squares (PLS) instead of an econometric fashion. methodology for treating heterogeneity in structural equa- 2. As one of the general importance-performance map analysis tion models. Marketing Science, 19(4), 328–347. https://doi. (IPMA) procedures, these path coefficients must be signifi- org/10.1287/mksc.19.4.328.11789 cant at any level of confidence before checking whether the Avkiran, N. K. (2018). Rise of the partial least squares struc- requirements for carrying out the IPMA have been established. tural equation modeling: An application in banking. In N. K. 3. According to Ringle and Sarstedt (2016, p. 1882), an indica- Avkiran & C. M. Ringle (Eds.), Partial least squares structural tor-related analysis is particularly useful in formative measure- equation modeling: Recent advances in banking and finance ment model settings. This work neglected the indicator-related (international series in operations research & management analysis because its indicators are reflective instead. science) (p. 239). Springer. 4. The previous work will be presented in this section due to its Avkiran, N. K., Ringle, C., & Low, R. (2015, July 16). Monitoring chronology. transmission of systemic risk from shadow banking to regulated 5. These requirements will be discussed later in this work (see banking. http://papers.ssrn.com/sol3/papers.cfm?abstract_id= section “IPMA.” 2634184 16 SAGE Open Avkiran, N. K., Ringle, C. M., & Low, R. (2018). Monitoring trans- Range Planning, 46(1–2), 1–12. doi:https://doi.org/10.1016/j. mission of systemic risk: Application of partial least squares lrp.2013.01.001 structural equation modeling in financial stress testing. Journal Hair, J. F., Sarstedt, M., & Ringle, R. M. (2018). Advanced issues in of Risk, 20(5), 83–115. partial least squares structural equation Modeling. Sage. Ayadurai, C., & Eskandari, R. (2018). Bank soundness: A PLS- Henseler, J., Ringle, C. M., & Sarstedt, M. (2015). A new criterion for SEM approach. In N. I. Avkiran & C. M. Ringle (Eds.), Partial assessing discriminant validity in variance-based structural equa- least squares structural equation modeling: Recent advances tion modeling. Journal of the Academy of Marketing Science, in banking and finance (pp. 31–52). Springer. 43(1), 115–135. https://doi.org/10.1007/s11747-014-0403-8 Barth, J. R., Joo, S., Kim, H., Lee, K. B., Maglic, S., & Shen, X. Henseler, J., Ringle, C. M., & Sarstedt, M. (2016). Testing mea- (2018). Forecasting net charge-off rates of banks: A PLS surement invariance of composites using partial least squares. approach. Department of Economics. http://cla.auburn.edu/ International Marketing Review, 33(3), 405–431. econwp/Archives/2018/2018-03.pdf Henseler, J., Ringle, C. M., & Sinkovics, R. R. (2009). The use Barth, M. E., & Landsman, W. R. (2010). How did financial report- of partial least squares path modeling in international market- ing contribute to the financial crisis? European Accounting ing. Advances in International Marketing, 20, 277–319. http:// Review, 19(3), 399–423. dx.doi.org/10.1108/S1474-7979(2009)0000020014 Beltratti, A., & Stulz, R. (2012). The credit crisis around the globe: Höck, C., Ringle, C. M., & Sarstedt, M. (2010). Management of Why did some banks perform better? Journal of Financial multi-purpose stadiums: Importance and performance measure- Economics, 105(1), 1–17. ment of service interfaces. International Journal of Services Cenfetelli, R. T., & Geneviève, B. (2009). Interpretation of Technology and Management, 14(2–3), 188–207. Formative Measurement in Information Systems Research. Jaafar, N. A., Noor, Z. M., & Mohamed, M. (2016). Student rat- MIS Quarterly, 33(4), 689–708. ings of teaching effectiveness: An Importance—Performance Chin, W. W. (1998). The partial least squares approach for structural Analysis (IPA). Journal of Educational and Social Research, equation modeling. In G. A. Marcoulides (Ed.), Methodology 6(3), 33–44. for business and management. Modern methods for business Joseph, M., Allbright, D., Stone, G., Sekhon, Y., & Tinson, J. (2005). research (pp. 295–336). Lawrence Erlbaum. Importance-performance analysis of UK and US bank customer Chin, W. W. (2010). How to write up and report PLS analyses. perceptions of service delivery technologies. International In W. W. Chin (Ed.), Handbook of partial least squares: Journal of Financial Services Management, 1(1), 66–88. Concepts, methods and applications (pp. 665–690). Springer. Kim, J., Chung, H., Amtmann, D., & Revicki, D. A. (2013). China, W. W., Mills, A. M., Steel, D. J., & Schwarz, A. (2012, May Measurement invariance of the PROMIS pain interference item 19-May 22). Multi-group invariance testing: An illustrative bank across community and clinical samples. Quality of Life comparison of PLS permutation and covariance-based SEM Research, 22, 502–507. http://dx.doi.org/10.1007/s11136-012- invariance analysis [Conference session] 7th International 0191-x Conference on Partial Least Squares and Related Methods, Kock, N. (2014). Advanced mediating effects tests, multi-group Houston, United States. analyses, and measurement model assessments in PLS-based Cox, R. A., Kimmel, R. K., & Wang, G. W. (2017). Drivers of US SEM. International Journal of e-Collaboration, 10(1), 1–13. bank failures during the financial crisis. International Journal Kock, N. (2015). One-tailed or two-tailed P values in PLS-SEM? of Business & Management, 12(8). https://doi.org/10.5539/ International Journal of e-Collaboration, 2(1), 1–7. ijbm.v12n8p19 Kock, N. (2017, June 2). Explore Measurement Invariance in Durvasula, S., Andrews, C. J., Lysonski, S., & Netemeyer, R. G. WarpPLS. WarpPLS. https://www.youtube.com/watch?v=29V (1993). Assessing the cross-national applicability of consumer qsAjhzqQ&feature=youtu.be behavior models: A model of attitude toward advertising in Kock, N., & Lynn, G. S. (2012). Lateral collinearity and misleading general. Journal of Consumer Research, 19(4), 626–636. results in variance-based SEM: An illustration and recommen- https://doi.org/10.1086/209327 dations. Journal of the Association for Information Systems, Erkens, D. H., Hung, M., & Matos, P. (2012, January). Corporate 13(7), 546–580. governance in the 2007–2008 financial crisis: Evidence Lee, L., Petter, S., Fayard, D., & Robinson, S. (2011). On the use from financial institutions worldwide. Journal of Corporate of partial least squares path modeling in accounting research. Finance, 18, 389–411. International Journal of Accounting Information Systems, Georgantopoulos, A. G., & Tsamis, A. D. (2013). Assessing the 14(2), 305–328. http://dx.doi.org/10.1016/j.accinf.2011.05.002 efficiency of commercial banks in Greece during the financial Lee, Y.-C., Yen, T.-M., & Tsai, C.-H. (2008). Using importance- crisis: A linear approach in conjunction with financial analysis. performance analysis and decision making trial and evaluation Journal of Money, Investment and Banking, 28, 31–46. laboratory to enhance order-winner criteria ~ a study of com- Hahn, C., Johnson, M. D., Herrmann, A., & Huber, F. (2002). puter industry. Information Technology Journal, 7(3), 396–408. Capturing customer heterogeneity using a finite mixture PLS Loureiro, S. M., & Miranda, F. J. (2011). Brand equity and brand loy- approach. Schmalenbach Business Review, 54, 243–269. alty in the Internet banking context: FIMIX-PLS market segmen- Hair, J. F., Hult, G. T., Ringle, C. M., & Sarstedt, M. (2016). A tation. Journal of Service Science and Management, 4, 476–485. primer on Partial Least Squares Structural Equation Modeling Lowry, P. B., & Gaskin, J. (2014). Partial Least Squares (PLS) (PLS-SEM) (2nd ed.). Sage. Structural Equation Modeling (SEM) for building and testing Hair, J. F., Ringle, C. M., & Sarstedt, M. (2013, February–April). behavioral causal theory: When to choose it and how to use Partial least squares structural equation modeling: Rigorous it. IEEE Transactions on Professional Communication, 57(2), applications, better results and higher acceptance. Long 123–146. Tailab 17 Maricica, M., & Georgeta, V. (2012). Business failure risk Sarstedt, M., Becker, J.-M., Ringle, C. M., & Schwaiger, M. analysis using financial ratios. Procedia – Social and (2011). Uncovering and treating unobserved heterogeneity Behavioral Sciences, 62, 728–732. https://doi.org/10.1016/j. with FIMIX-PLS: Which model selection criterion provides sbspro.2012.09.123 an appropriate number of segments? Schmalenbach Business Martilla, J. A., & James, J. C. (1977). Importance-performance Review, 63(1), 34–62. analysis. Journal of Marketing, 41(1), 77–79. Sarstedt, M., Ringle, C. M., & Hair, J. F. (2018). Treating unob- Ngah, R., Ramayah, T., & Sarmidy, R. (2018). Partial least square served heterogeneity in PLS-SEM: A multi-method approach. analysis on micro enterprises’ intellectual capital and per- In H. Latan & R. Noonan (Eds.), Partial least squares path formance: The mediating effect of tacit knowledge sharing. modeling: Basic concepts, methodological Issues and Journal of Applied Structural Equation Modeling, 2(2), 22–33. Applications (pp. 197–213). Springer. Nitzl, C. (2016). The use of partial least squares structural equa- Schlagel, C., & Sarstedt, M. (2016). Assessing the measurement tion modelling (PLS-SEM) in management accounting invariance of the four-dimensional cultural intelligence scale research: Directions for future theory development. Journal of across countries: A composite model approach. European Accounting Literature, 37, 19–35. http://dx.doi.org/10.1016/j. Management Journal, 34, 633–649. acclit.2016.09.003 Serrano-Cinca, C., Fuertes-Callén, Y., Gutiérrez-Nieto, B., & Ramayah, T., Chiun, L., Rouibah, K., & May, O. (2014). Cuellar-Fernández, B. (2014). Path modelling to bank- Identifying priority using an Importance-Performance Matrix ruptcy: Causes and symptoms of the banking crisis. Applied Analysis (IPMA): The case of Internet banking in Malaysia. Economics, 46(31), 3798–3811. International Journal of E-Adoption, 6, 1–15. https://doi. Serrano-Cinca, C., & Gutiérrez-Nieto, B. (2013). Partial Least org/10.4018/ijea.2014010101 Square Discriminant Analysis for bankruptcy prediction. Rasoolimanesh, S. M., Ringle, C. M., Jaafar, M., & Ramayah, T. Decision Support Systems, 54, 1245–1255. (2017). Urban vs. rural destinations: Residents’ perceptions, Silva, F., & Fernandes, F.O. (2011). Importance-Performance community participation and support for tourism development. Analysis as a Tool in Evaluating Higher Education Service Tourism Management, 60, 147–158. https://doi.org/10.1016/j. Quality: The Empirical Results of ESTiG (IPB). Creating tourman.2016.11.019 Global Competitive Economies: A 360-Degree Approach Ringle, C. M., & Sarstedt, M. (2016). Gain more insight from your Streukens, S., Leroi-Werelds, S., & Willems, K. (2018). Dealing PLS-SEM results The importance-performance map analysis. with nonlinearity in Importance-Performance Map Analysis Industrial Management & Data Systems, 116(9), 1865–1886. (IPMA): An integrative framework in a PLS-SEM context. In https://doi.org/10.1108/IMDS-10-2015-0449 H. Latan & R. Noonan (Eds.), Partial least squares path mod- Ringle, C. M., Sarstedt, M., & Mooi, E. A. (2010). Response- eling: Basic concepts, methodological issues and applications based segmentation using finite mixture partial least squares: (p. 452), Springer. Theoretical foundations and an application to American cus- Wyród-Wróbel, J., & Biesok, G. (2017). Decision making on vari- tomer satisfaction index data. In R. Stahlbock, S. F. Crone, & ous approaches to Importance-Performance Analysis (IPA). S. Lessmann (Eds.), Data mining: Annals of information sys- Faculty of Business and Economics, 3(2), 123–131. https://doi. tems (vol. 8, pp. 19–49). New York: Springer. org/10.11118/ejobsat.v3i2.82 Ringle, C. M., Wende, S., Will, A. (2010). Finite mixture partial Zaghdoudi, T. (2013). Bank failure prediction with logistic regres- least squares analysis: Methodology and numerical examples. sion. International Journal of Economics and Financial Issues, In V. Esposito Vinzi, W. Chin, J. Henseler, H. Wang (Eds.), 3(2), 537–543. Handbook of partial least squares. Springer handbooks of com- putational statistics (pp. 195–218). Berlin, Heidelberg: Springer. Author Biography Ruzzier, M. K., Antoncic, B., & Ruzzier, M. (2014). Cross-cultural model of customer-based brand equity for a tourism destina- Mohamed M. Khalifa Tailab is an assistant professor of busi- tion. IUP Journal of Brand Management, XI, 7–29. ness administration and accounting at Lincoln University. His Samar, S., Ghani, M. A., & Alnaser, F. (2017). Predicting cus- primary research interests include corporate disclosure, textual tomer’s intentions to use internet banking: The role of tech- analysis modeling, market reaction, and the application of partial nology acceptance. Management Science Letters, 7(11), least squares structural equation modeling (PLS-SEM) in busi- 513–524. ness research.

Journal

SAGE Open – SAGE

Published: Jan 26, 2020

Keywords: partial least squares (PLS); structural equation modeling (SEM); importance-performance map analysis (IPMA); banking crisis; measurement invariance; multiple-group analysis

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Using Importance-Performance Matrix Analysis to Evaluate the Financial Performance of American Banks During the Financial Crisis:

Using Importance-Performance Matrix Analysis to Evaluate the Financial Performance of American Banks During the Financial Crisis:

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Using Importance-Performance Matrix Analysis to Evaluate the Financial Performance of American Banks During the Financial Crisis:

Using Importance-Performance Matrix Analysis to Evaluate the Financial Performance of American Banks During the Financial Crisis:

References (50)

Abstract

Journal

Recommended Articles

There are no references for this article.

Our policy towards the use of cookies