Empirical studies on the performance of banks: A systematic literature review for future research

This paper intends to review research on the performance of banks to identify gaps in the current body of knowledge to justify future research directions. We use a systematic literature review method and review 164 articles from refereed journals. Content analysis reveals that most of the studies are empirical focusing on two aspects i.e. financial performance and efficiency of banks. These studies consider the impact of particular events and contexts on performance and efficiency while testing research hypotheses. However, often there is a lack of a theoretical backing for these studies. We argue that the considered events and contexts affect the risk transformation process under the financial intermediation theory. The efficiency of banks reflects the risk transformation process and causes performance. On the other hand, traditional performance indicators were based on financial measures that do not reflect the components of the risk transformation process. A sound comprehensive risk-based composite measure is required to fill this gap.


Introduction
A function of a bank differs from other business firms. The main economic functions of the banks are those of consolidating and transforming risks on the one hand by providing credits in the deficit sector and reducing the cost of transaction and information costs on the other hand. The theory of banking firm explains the risk transformation process. In this context, the performance of banks is a very popular research area over the years which focuses on the impact of a selected context or an event on the performance. These studies cover the banking process as well as the management issues of banks. The banking process has been discussed in different theoretical perspectives. However, most performance studies have been empirical studies that lack a sound theoretical foundation on the backing process. Three theories have emerged over time in the banking process i.e. credit creation theory, fractional reserve theory, and financial intermediation theory. Today, financial intermediation theory is dominant in the banking process. Some other theories are related to the management issues of banks i.e. agency theory, stakeholder theory, trade-off theory, resource-based theory, moral hazard theory, and capital buffer theory, among others. We will focus on the performance of banks under the banking process in terms of financial intermediation theory.
Empirical studies show that performance measures depend on the objectives of the researcher. There is no unique measure of performance and therefore, the same study has used several measures to capture the performance. However, unique results among different measures have not been generated. Therefore, this study intends to review the empirical work to identify a suitable method of measuring the performance of banks in different research perspectives or context. Therefore, the purpose of this study is to provide an extensive literature review on the performance of banks. More specifically, the objectives of the study are to: 1. Synthesize the existing literature on the performance of banks. 2. Arrange the publications in an orderly manner to enable easy and quick search. 3. Classify performance research articles according to their approach and methodologies. 4. Explore the issues in performance research and suggest research areas for future work.
This study also focuses on the following research questions in addition to the above objectives. These questions will guide us to have a comprehensive understanding of the research context.

•
How articles on bank performance are placed in time?

•
What has been studied about performance? • What research methods have been used? • What are the most important articles about the topic?
The remainder of the paper is organized as follows. We first introduce the concept of performance of banks and research methodology. Second, we provide an article classification scheme. Thereafter, an overview of the selected publications is provided. Finally, we outline our main findings, discuss implications, and suggest avenues for future research.

Concept of Performance of Banks
Cambridge Dictionary defines performance as "how well a person, machine, etc. does a piece of work or activity". According to the Oxford dictionary, performance is the action or process of performing a task or function. We define the performance as how well a bank does the financial intermediation process in sustainably achieving the objectives of stakeholders. The performance of banks is assessed in different perspectives using financial as well as non-financial indicators of banks. Hughes & Mester (2015) identify two broad approaches in measuring the performance of banks i.e. non-structural and structural approaches. Non-structural approaches use different performance measures (e.g. ROE, ROA, net interest margins, Tobin's q-ratio among others). In contrast, structural approaches are based on theoretical models of banking behavior such as efficient and profit frontiers. The efficiency measures how best a bank performs over the other banks in the industry in the process of converting inputs into outputs. Banks are different from the other business entities and they function as an intermediary between depositors and borrowers. Therefore, the performance of banks has to be measured considering the financial intermediation process. The overall long-term financial performance results in the financial stability of banks.

The Methodology of the Literature Review
A systematic approach was used to identify and select articles for the review. Fink (2005) suggests that a literature review must use a methodology that is systematic in approach, explicit in describing the procedures by which it was conducted and comprehensive in its scope of including all relevant material related to particular phenomena. Following Singh & Kumar (2014), a four-category process was used as shown in Figure 1.

Literature Collection and Boundary Identification
We searched for articles in Elsevier and Emerald databases to capture a comprehensive and diverse range of articles published in Elsevier and Emerald databases for review. We started with a keyword search and then delimited the literature thereafter. Thus, it is a combination of deductive and inductive approaches (Singh & Kumar, 2014;Shukla & Jharkharia, 2013). The keyword search generated a sizable number of articles. We adopted the methodology used by Singh & Kumar (2014) to filter the most relevant literature. The boundaries for selecting papers were given as follows: • Papers published only in peer-reviewed indexed academic journals were considered. • Papers were collected for 13 years (2008-2020).

•
Papers with full-text available were considered.
This literature search was carried out considering the keywords within the boundaries to identify a sample for further screening. We used the performance of banks as the keyword and checked in titles, keywords, or abstracts for the identification. The first round search generated 795 articles within delimiting boundaries. Subsequently, we filtered the articles to identify the most relevant articles for our research context. This filtering process was carried out by reading the identified papers considering the relevance and appropriateness. The detailed comprehensive analysis produced 164 articles and was qualified for the final review. The protocol for the database search is illustrated in Table 1. The summary of the systematic process of article selection is presented in Figure 2. We searched Elsevier and Emerald databases from 2008 to 2020 on 19.11.2019.

Description and Classification of Literature
The identified 164 were analyzed through the systemic literature review concerning the publication year, coverage of the study period, the methodology used, journal of publication, types of banking sector studied, country of studies, and source of data to understand the trends and issues in the literature.

Analysis by Year of Publication
This analysis intends to identify significant contexts and events through the dispersion of articles over the sample period. The analysis shows that the selected articles have been published quite Step 1: Data identification and article selection Step 2: Description and clasification of literature Step 3: Detailed content and citation analysis of articles Step 4 : Reporting of findings and research gap Journal of Research in Emerging Markets, 2020, 2(4).

Analysis by Research Methods
The analysis reveals that all the papers were empirical studies. Table 2 shows the frequency of different empirical methodologies of our sample. We can identify three categories of empirical studies which include empirical panel data, OLS regression, and other measures of efficiency including Data Envelopment Analysis (DEA), Stochastic Frontier Approach (SFA), and cost efficiency measures among others. The panel data analysis has been the most popular methodology which represents 59.76 percent of total studies. Under the panel data, the majority has focused on the system generalized method of the moment. On the other hand, only forty-two studies are based on other efficiency-based performance measures.

Analysis by the journal of publication
This analysis is used to identify the most influential and involved journals which may be useful for future researchers and also will be useful to continue the conversation on the performance of banks. Our sample was contained in 59 journals as shown in Figure 5. Journal of Banking and Finance provides a maximum of twenty articles, followed by the Journal of Financial Stability with thirteen articles.

Analysis by Type of Banks
The performance studies cover all the subgroups in the banking industry. However, there are some specific studies on commercial banks, investment banks, savings banks. And also, the different types of characteristics such as government, private, and foreign ownership, or listed in stock exchange are considered. Our review suggests that commercial banks are dominant in performance studies. However, recent articles show a trend of studying the performance of Islamic banks concerning the performance of conventional banks (Majid, Falahaty, & Jusoh, 2017;Alam, Zainuddin, & Rizvi, 2019;Azad, Azmat, & Hayat, 2019;Berger, Boubakri, Guedhami, & Li, 2019). The majority of articles are cross-country studies which indicates an attempt to compare outcomes and findings across countries. Our analysis reveals that there are differences in performance among different categories of banks. For example, Köhler(2015) shows that listed banks have a significantly higher share of non-interest income and they are also more dependent on non-deposit funding than banks that are not listed. This might reduce the benefit of diversifying into non-interest income and non-deposit funding. Beck, Demirgüç-Kunt, & Merrouche (2013) show the better stock performance of listed Islamic banks due to their higher capitalization and better asset quality over the conventional banks. Zuhroh, Ismail, & Maskie (2015) show that Islamic banks are efficient over the conventional banks. Table 3 shows the frequency of articles for each subgroup of banks.

Analysis by Country of Studies
The analysis by country of studies provides some peculiar insights from the literature. Figure 6 provides the mapping of countries based on the coverage of the number of studies. This shows that performance studies are concentrated in East Asian count and Eastern European countries over the other regions. We assume that the East Asian financial crises and the membership of the European Union by Eastern European countries have fueled the interest. Apart from these two regions, China, India, and the USA have been dominant in performance studies which may be due to the size and growth of economies where banks play a major role. It is a common observation that there is a trend of focusing on a group of countries or regions for the studies rather than concentrating on a single country. There is more concentration on Asia over other regions. Malaysia, Turkey, Indonesia, China, and India are the frequently studied countries during the period of the survey. A list of countries with ten or more research articles is presented in Figure 7.

Analysis by Source of Data
The performance studies considered in this review had used secondary data to gather bank-level as well as country-level data. The bank Scope database was popular in obtaining bank-level data while the world development indicators compiled by the World Bank and International Financial Statistics database compiled by the International Monetary Fund were popular for the macroeconomic data. The frequently used sources for bank-level data are shown in Table 4.

Citation Analysis
A citation is a reference to the work of another author/s in someone's academic work. Citation analysis is the study of cited references for a population of articles to find the most influential works in the field (Singh & Kumar, 2014). Google Scholar is used to capturing citation information of the sample. Our analysis found that only 141 of the 164 articles had been cited. Most of the uncited articles have been published recently during 2018-2020. The cited 141 articles count 12,758 cited references at an average of 80 per article. The most-cited articles with at least 100 citations are shown in Table 5. (Athanasoglou, Brissimis, & Delis (2008) is the most-cited article with 1474 citations. Andres & Vallelado (2008) Berger & Bouwman (2013), and Demirguc-Kunt & Huizinga (2010) were also among the top five most-cited publications. Table 6 presents the most referenced journals that published work on the performance of banks. In social sciences, academic accreditation is recognized as the publications in ranked journal databases i.e. ABDC, ABS, Scopus, and ISI. The social science academic community requires the ranked journal in ABDC & ABS to be included in the Scopus index

Content Analysis
Content analysis is a class of methods within social science that can be applied both in quantitative and qualitative ways (Singh & Kumar, 2014). The content analysis is used to see what the other authors have done in their studies. Detailed content analysis of our sample reveals that most of the studies on the performance of banks have been focused on the financial performance and efficiency of banks considering determinant factors. These studies have not considered the financial intermediation process in performance measurements. Most of the studies focus on the impact of a given event, scenario, or context on the performance and efficiency of banks. For example, the impact of capital on bank performance during financial crises; bank bailouts and bank performance; non-performing loans and bank efficiency; bank political connections and performance; bank reforms, foreign ownership, and financial stability; bank regulation and efficiency; bank regulations, competition, and financial reforms on banks' performance; bank supervision, regulation, and efficiency; banking sector reform and performance; board diversity and its effects on bank performance; board structure and performance; CEO education and bank performance; corporate governance and bank performance; corporate social responsibility and financial performance; determinants of bank performance and bank efficiency; were popular among other areas. We propose that the above events and contexts affect the intermediation process that can be measured in terms of efficiency. This efficiency causes the shortterm performance of banks which is measured by using accounting ratios and long-term performance in terms of stability.  (2018) ROA, Capital ratio Bian & Dend (2017) ROA, ROE, NPL King, Srivastav, & Williams (2016) Industry adjusted ROA Musali & Ismail (2014) ROA, ROE Maqbool & Zameer (2018) A composite index of performance (ROE, ROA, NP, SR, and PE) Rashid & Jabeen (2016) A financial performance index using CAMELS variables.
The studies have considered two determinant factors on the performance of banks .i.e. internal and external factors. According to Robin, Salim, & Bloch (2018), the internal determinants are related to bank-specific variables. Commonly used internal factors in determining the financial performance of banks are shown in the following table.

Efficiency of Banks
The efficiency is another approach that is used to measure the performance. The efficiency measures how best a bank performs over the other banks in the industry in the process of converting inputs into outputs. There are various approaches to measure the efficiency of banks (see Table 9).
Studies on the efficiency of banks also have considered two types of determinants i.e.: internal and external (Manthos & Nikolaos,2009;Mansour & Moussawi, 2019). The internal factors are bankspecific and external factors are related to the external environment or the macroeconomy.

Lack of Theoretical Backing For The Studies
Most empirical studies are not backed by theories. Some papers link the results with resourcebased theories considering capital and other resources. Banks functions as intermediaries between depositors and borrowers which involve a risk transformation under the financial intermediation theory. Therefore, the performance of banks has to be studied in terms of the financial intermediation process. The previous studies lack a systematic combination of the empirical results of the performance of banks with the financial intermediation theory. We can identify two reasons for this limitation. First, the researchers have tried to link the impact of events or context with the performance. Secondly, researchers have focused on testing the hypotheses related to the performance and profitability of banks. This background provides an opportunity for extending the financial intermediation theory. We argue that the events and contexts affect the risk transformation process which can be measured in terms of efficiency under the financial intermediation theory. This improvement or efficiency of the process affects the short-term performance of the banks. Sustainable continuous short-term performance affects the financial stability of banks. Therefore, modified financial intermediation theory considering the risk transformation process will take the following form.

Limitations Of Risk Components in Performance Measures
The most of considered papers had considered accounting ratios to measure the performance of banks i.e. return on assets, return on equity, net interest margin, among other measures. The Z-score also has been considered as a measure of performance. We believe accounting ratios measure the shortterm financial performance of banks and Z-score measures the long-term performance of banks. Therefore, we are of the view that the performance of banks has to be measured considering the risk transformation process of banks using a composite index to capture all the risks. The weights for the variables have to be decided on the respective weights of the components represent in the balance sheet of banks. There were some initiatives of developing such indices. Maqbool & Zameer (2018 develop a composite performance index but it does not consider the risk components. Rashid & Jabeen (2016) develop a financial performance index using CAMELS variables. However, they give equal weights for the selected variables which do not reflect the actual risk levels of banks. Our analysis focuses on studies of performance measurement. However, measurement error is not the only major challenge to the empirical studies of bank performance. For example, the study of performance is subject to endogeneity problems, which include not only measurement errors but also other issues such as omitted confounding factors, unobserved heterogeneities, and simultaneity, among others.

Dominance of Empirical Research
The review shows that all the research articles were based on secondary data. There is a dearth of empirical studies in African and some other emerging countries. This may be due to the cost involved in accessing bank-level data from databases as well as limitations of disclosure requirements in underdeveloped capital markets.

Limited Studies in Emerging Countries
There were only 85 single-country studies of which 25 were in emerging countries. The remaining 79 articles were inter country or international studies.

Conclusion
The main objective of this literature review was to identify and analyze the literature on the performance of banks to understand the research context in-depth to contribute theoretical development in bank risk transformation process. We identified and analyzed 164 research articles published during 2008-2020 on the performance of banks by adopting a systemic literature review methodology. There is a trend of increasing the number of research articles over time. However, there is a limitation of linking performance with the financial intermediation process. The major focus of performance studies was to determine the financial performance and efficiency of banks based on a given event or context. However, researchers were reluctant to adopting risk-based performance measures to capture the risk transformation process in financial intermediation. Therefore, in summary from the complete discussion, it is concluded that future research should focus on extending financial intermediation theory considering the effect of context and events on the intermediation process and performance of banks.
This literature review on performance measures of banks has highlighted conceptual and methodological issues which may serve as guidance for future research

Research question
Main conclusion How articles are placed in time?
Publication of articles shows an increasing trend over time which may be due to the continuous development in the banking industry in different dimensions giving events and context for researchers to study the performance. Further, the financial crisis and recent economical slow down have fuelled the performance studies. What has been studied about the performance of banks?
Most of the studies were focused on studying determinants of financial performance or efficiency of banks based on internal and external factors in a given event or a context. What research methods and data have been used?
Most often panel data have been used for the studies and data have been collected from databases as well as published annual reports of the banks. What are the most important articles about the topic?
The most popular articles in this area are (Athanasoglou, Brissimis, & Delis(2008), Andres & Vallelado(2008) (2010), and Kosmidou (2008). However, their popularity is largely caused by the fact that they are early articles that studied the performance of banks after a major context or an event such as a financial crisis.
We considered only the banking process. However, management related theories are also important in this kind of analysis. It is suggested to consider the following areas for future research: • One such avenue warranting research attention is the impact of the behavior of the management of banks as we proposed in figure 8. Existing research literature focuses on the impact of management and various determinants on the performance of banks but left the risk transformation process in the banking aspect.

•
Future research can also be directed toward the establishment of banking sector-specific measures for measuring performance in different segments i.e. conventional vs. Islamic, commercial vs. investment, savings vs. development, among other classifications, adopts different banking practices in the intermediation process, hence there may be an impact on performance. Further, future studies may focus on interdisciplinary perspectives in performance studies. Besides, to focus on more important external factors such as market risk, competition and market power, and concentration. Therefore, the formulation of the conceptual framework required. Funding: .This research received no external funding.