The impact of Internet Banking on Italian and Romanian Bank Performance

The impact of Internet Banking on Italian and

Romanian Bank Performance

Abstract:

Internet banking is acquiring more importance in the banking industry due to the evolution of technology and the consumers’ needs. This paper examines the relation between the adoption of Internet Banking and the profitability measures in a sample of 9 Romanian banks and 28 Italian banks between 2003 and 2013. Background theory provides mixed results and further research, which the paper tries to address. The research shows that the adoption of Internet Banking does not increase the profitability measures of banks, but it affects Cost-to-Income ratio negatively,

providing more efficiency to the banks. Future research may investigate the impact of Internet banking on bank’s performance using a different sample.

Key words: bank performance, internet banking, cost-to-income ratio Student number: S3047636

Name: Bianca Serban

1

1. Introduction

The traditional banks that were established hundreds of years ago have evolved and they have changed most of their original characteristics. The banking system involves the entire global economic system and it is a reflection of the real situation of a country. Developments in technology have spread also the bank department, creating a new service: Internet banking or e-banking. It has assumed different names and different connotations but the benefits that derived from it are equal in all the types of banks. The digital transformation process has spread in all the sectors of the society, affecting also the banking system. Institutions, from the largest to the smallest, have understood how important it is to adapt to technological innovations.

Internet banking is a channel for marketing, it ensures consumer satisfaction through new services and it allows implementing services never used previously. Increasing customer satisfaction, the prestige and the efficiency of the bank also increase. On the other side, Internet banking may be disadvantageous because has a high cost to maintain. Regarding the customers’ needs, Internet banking may be a channel that leads to lose the contact between the client and the real bank. There are several advantages and disadvantages for a bank that is involved in the decision process for implementing Internet banking. Analyzing the effects that the multichannel service has on the

measures of a bank, may lead to eliminate some advantages or to add new ones. The research question of this paper wants to investigate whether the adoption of Internet bank leads to a higher level of profitability. Romania is one of the two countries analyzed and is the principal component of the paper because the Romanian banking system moved from a state-bank model to a market-bank model in the last years, precisely in 1998. Completely different institutional

backgrounds are presented for the bank’s home countries Romania and Italy. Moreover, Internet banking was implemented in more recent years in Romania, compared to Italy. CEC bank, a Romanian bank from the sample, implemented Internet banking in 2008. Another important element to be considered is the late entry of Romania in the European Union (2007). On the

contrary, Italy is a founding member of the European Union. This explains the different institutional background that exist between the two countries before 2007. In fact, Romania, in the time of accession to the European Union, acquires similar directives acquired by Italy.

2

banks) over the period 2003-2013. The data concerning the sample are taken from the database Bankscope. The data concerning the implementation of Internet banking in Romania are extrapolated from the paper “The impact of internet banking on the performance of Romanian banks: DEA and PCA approach” (Stoica et al., 2013). On the other hand, the data regarding the implementation of e-banking in Italy were obtained through an accurate study on the specific web sites of the banks. Some banks do not share this type of data and therefore information were required by direct e-mail. The model applied to the set of regressions is based on the study of Al-Sa’ Di and Khwarish (2011) and the equation considers Return on Assets, Return on Equity and Cost to Income ratio as dependent variables. The latter are subject to variations due to the

implementation of Internet banking. Moreover, a comparison between the results obtained in the two countries is examined in order to

have a clear understanding of the difference in Internet banking between Italy and Romania. Furthermore, the international component of this research leads to create two sub-sample, one for the Romanian banks and the other one for the Italian banks, underling the same research question: “are performance measures positively correlated with the adoption of Internet banking?”. Prior research on this topic has been made, in particular Hernando and Nieto (2006), Hasan et al (2008), Arnaboldi et al. (2008) and Al-Sa’Di and Khwarish (2011). These countries are different for institutional backgrounds, culture and banking industry. Certainly, a country advanced in the field of banks and Internet use is Italy. Conversely, even in a geographical way, it is Romania. Italy has very remote roots regarding the banking system and being a developed country, it is also considered very fast in adapting to technological changes. The economic and political background of the two countries leads to a different perspective regarding the implementation of Internet banking and its impact on banks’ measures.

The evolution of the banking sector in digitalizing market can be summarized into three models. First, pure digital model that do not require human interaction, it deals with specific issues such as payments, transactions, loans, remittances of money, and even investments. Second, multichannel digital model prevalence, in which coexists a strong digital component and a human support that deals with customer services and sales. Third, multi-channel model in human prevalence, used mainly by traditional banks with strong ties to the territory. These three models coexist and share the same target customers. In order to be competitive, banks have to pay attention to challengers’ offers. Internet banking is built on the ability of providing customers with certain functionality,

targeted to their specific needs. Internet banking has revolutionized the banking industry and it is no longer considered as a product,

3

rapid pace and the first initiators were the larger banks. It has become the symbol of each bank and it is advertised everywhere. Indeed, also the least developed countries, with a weak banking system, have tried to keep up with the evolution of internet banking. This does not mean that the adoption has been successful everywhere. Even nowadays, many banks are skeptical about the initial costs of implementation and their customers are in doubt about the safety of the service. In fact, many researches are done regarding the risks related to a possible adoption of Internet technology (Hasan et al., 2008). All of the largest European and US banks offer Internet banking options and even the smallest banks try to stay in time with this process. A recent KPMG survey on banking (2016) reported that banks with smaller assets sizes have more ambitious plans to increase digital products than their larger counterparts. Such change does not come easy, nearly one-third of banking

executives believe changes is always included at their institution and 44% is often included. The negative factors that are taken into consideration at the beginning are especially fear and unclear benefits. Furthermore, due to the complicated global economic and financial framework, the branch transformation has become essential and resulted in a real reduction in the number of branches within the various banking groups. Indeed, in October 2016 there was an important business news regarding this topic. The credit sector is going through a period of profound changes, which affects the balance of many institutions and also the business model: less branches, more web, and new relationship with clients. ING Group, the largest Dutch bank, has analyzed the possibility to cutting capacity of 7,000 jobs, mainly in Belgium and the Netherlands in order to face the program of a costs’ reduction of 900 million euro by 2021. In addition, ING announced a massive investment of 800 million over the next five years in a constant transformation towards digital in order to further improve the customer experience, accelerating growth in the number of customers and loans

(Repubblica.it, 2016). Lastly, a study commissioned by Wincor Nixdorf to IDC shows that in 2017, banks worldwide will invest about 16 billion dollars in the conversion of branches and in the technological adaptation necessary to achieve it (Wincor-Nixdorf, 2016).

This study is structured as follows: First section is containing the introduction of Internet banking concerning the motivation of the research. Chapter two contains the literature review on the

4

research of the study. In the end, the references used for the research are presented, followed by the appendices.

2. Literature Review

This section elaborates on the current literature. It gives an overview of the background papers. Impact of Internet on banks’ performance has been studied for multiple countries and in several periods of time. Although it is a recent development, scholars were analyzing the effects that Internet banking can have on banks, on consumers, on market. The first implementation of Internet banking and the research of its effects is in the United States. Over time, Internet banking has also expanded across the European banks and the literature investigates its impact and the difference between the European effects and the US effects.

The implementation of Internet banking makes a difference among the types of banks and the services they offer. The comparison is between “brick-and-mortar” banks and “click-and-mortar” banks (DeYoung et al., 2006). The first are traditional banks with several branches and all the services to be performed in the office. In fact, they not offer the customers the flexibility in

managing operations due to the opening hours of the branches. In contrast, the Internet bank allows transactions at any time of day and it offers economic incentives as more competitive interest rates on deposits and loans (Tsai et al., 2013). The traditional distributional channels as branches, ATMs, and telephone are used in the multichannel banks together with the innovative product of Internet banking. They offer economic incentives (more competitive interest rates on deposits and loans, lower commission fee for their banking services) to their customers in order to obtain additional clients in the new distribution channel (Hernando et al., 2006). Internet banking was defined as the virtual interaction between account holders and banks, resulting in financial services (Zolait et al., 2010). It is a business model in which a bank integrates both offline (bricks) and online (clicks) presences (Cyree et al., 2009). Furthermore, Internet banking was able to eliminate the creation of new branches, excluding the overhead expenses of conventional banks and service customers (Zolait et al., 2010). The Internet bank channel offers opportunities such as low-cost banking, profitable banking, quality banking and allows banks to sell services customized (Lu, 2005; Sahut, 2009). In fact, this paper seeks to determine the Impact that Internet banking adoption has on the performance measures and according to Lu (2005) and Sahut (2009), the new channel leads to a higher profitability. In order to understand the relationship between performance ratio and Internet, it is important to explain what kind of services and opportunities Internet banking offers to

5

payment via website, finding mortgages or auto loans, applying for credit card (Mattila et al., 2003). The banking opportunities offered by internet services are convenient and precise and the products are designed to the customer’s individual need (Dandapani et al., 2008). The customer’s individual need is the more important element that Internet banking contains. In fact, the introduction of the new service connects to a novelty and convenience of consumer’s needs.

The advantages of Internet banking can be divided into three classes. The latter include the

beneficiaries of the implementation of Internet banking. The advantages for the bank are: improved market image with the new product, that lead to become leader in the field; driving away

competitors; reducing transaction costs; decreasing the number of workers; Internet sites and platform to advertise the new financial products; market expansion because the new channel is accessible from all over the world; and quick response to the evolution of technology and to the banking market demand. The advantages for the individual client and for the institutional client are: quicker and easier transactions, the service can be used 24 hours a day, without requiring the

physical interaction with the bank; lower costs in accessing and using the banking services; all transactions are recorded and accessible at all times. The major disadvantage in using Internet banking is the security for the individual client. The client’s code used to access the media bank can be hacked or the telephone used to make transactions can be stolen.

2.1 Relationship between Internet banking and bank performance in non European countries

In 1998, Egland (1998) et al. investigated the relationship between Internet banking and

performance of banks in the US. They analyzed the structure and performance characteristics of these banks but they did not find fundamental differences between the performance of Internet banks and non-Internet banks. Sullivan (2000) examined the transactional web sites of the multichannel banks in the 10th _{Federal Reserve District and in particular the reluctance of small}

banks in adopting the new method. He argued that larger non-Internet banks hold more business loans compared to Internet banks. The latter, in particular the large regional banks, tend to present high non-interest expenses but profitability resulted equal for both types of banks. They used a t-test in order to determine the relationship between the performance indicators and a sample of Internet banks and also the association between non Internet banks and profitability measure. The period

taken in consideration was between the second quarter of 1998 and the third quarter of 1999. Furst et al. (2002) calculated the performance of Internet banks, mainly in terms of ROE in 1999,

6

adoption rates, with the larger banks the more likely to adopt internet banking; that Internet banks create their income from non-interest sources and use federal funds in order to finance their assets; that credit quality and exam ratings are slightly better amid Internet banks. Regarding the research question of this paper, the sample contains small banks concerning the Romanian sample and this might be a reason for the late adopting of the multichannel. Carlson et al. (2002) analyzed the possible link between Internet banking and bank’s profitability in the US market. They ran a logit regression model estimating a bank’s return on equity (ROE) against a set of control variables, including an explanatory variable that notes the presence or not of Internet banking and they have compared this relation. In doing so, they create a dummy variable that explains Internet offered or not and put in a relationship with the ROE value in a certain period

of time, 1999. The results of the regression presented no impact of Internet on the profitability. Sathye (2005) proposed Internet banking as a performance-enhancing tool for the 61 credit unions

in Australia between 1997 and 2001. From the sample of 61 credit unions, 44 were recognized as IECUs, Internet experienced cred union. He used DEA in order to measure the performance of the credit unions and for the Internet adoption, he used IECU as a variable equal to 1 if the credit union adopted Internet in 2001. His results reject the proposal and show that Internet banking neither reduces nor enhances risk profile.

All these initial studies have had similar findings. There were no systematic evidences that banks were helped or harmed by offering the innovative Internet banking. In particular, the performance of banks is not higher with the adoption of Internet banking.

In 2006, DeYoung et al. observed 424 community banks, chosen between the first US banks in adopting Internet banking in 1990. The period of the sample studied is between 1999 and 2001. DeYoung et al (2006) compared the 424 community banks with other 5175 traditional community banks on the financial performance ratios. They find a positive relationship between the

communities bank and the profitability ratios.

With the passing years, the authors have also focused on smaller banks and especially on European banks. In fact, the adoption of internet banking came later in Europe.

2.4 Relationship between Internet banking and bank performance in European countries

7

technology-based scale economies arising from their ability to control operational expenses more efficiently than the traditional new banks. On the contrary, traditional banks can increase lending activity as they gain experience, something that Internet banks cannot. Hernando and Nieto (2005) analyzed a sample of 72 commercial Spanish banks, in the period between 1994 and 2002. They estimated an increase in the performance of the banks with the adoption of Internet. The variables used for the performance were Return on Assets and Return on Equity. Moreover, in their study, Internet adoption is significant in reducing overhead expenses, and increasing brokerage fees and commission income. After three years, in 2008, Onay, Ozsoz and Helvacioglu studied the relationship between internet banking and bank performance. The sample represented 13 banks that used Internet banking during the studied period, in Turkey. The period for the sample was between 1996 and 2005. The variables taken in consideration for the profitability of the banks were Return on Assets and Return on Equity, using the methodology of Hernando and Nieto (2005). They also analyzed the relationship using macroeconomic control variables. The results were in line with the paper used for the methodology, Return on Assets was positively affected by the adoption of Internet banking.

Bonaccorsi di Patti et al. (2004) analyzed the relevance of demand-side complementary between electronical and traditional provision of banking services. The authors did not find any benefit in using the multichannel for Italian banks between 1998 and 2001. The evidence suggested that banks did not benefit from a consumer surplus from the joint provision of online banking services and traditional services provided at the branches. Conversely, Hasan et al. (2008) investigated the impact of Internet and innovative banking products on banks’ performance. The authors suggested a positive link between Internet adoption and bank profitability (ROA, ROE) in Italy between 1993 and 2001 period. In conclusion, bank returns are positively correlated with banks’ performance, whereas the adoption of Internet activities is negatively correlated with bank risk. De Young et al. (2006) suggested an increase in bank profitability as a consequence of adding

Internet channel to an existing multitude of bank branches. Noninterest income extracted from deposit service charges is the main factor of the increase in profitability. Dandapani et al. (2008) found that Internet banking could increase a bank’s assets growth but it is also able to increase operating costs. In fact, the additional customers acquired from internet banking services could generate greater interest rate spreads but the difficulty in attracting new customers will be considered a cost.

8

internet banking and if there were differences or similarities in mixed banks that use online banking, and pure online banks. The Panel data used was from 1995-2004 year with 46 banks. The authors found no important difference between online banks and banks with internet banking and in conclusion it was observed an increase of the IT departments costs and personnel and that the management of the banks were more oriented towards the online banking offering.The study by Lambrecht and Seim (2006) analyzed German banks, the relative consumers adoption and usage of online banking from August 2001 till July 2003. The final conclusion is that authors found the divergence between customer adoption in online banking whether consumers adopt and when they adopt online banking. In fact the online banking adoption is very much related to educated people, with higher income and mostly male.

In sum, the opinions on this topic are contradictory and are still in discussion. 2.5 Internet Banking in Italy

The first web site of an Italian bank has been created by the Cassa di Risparmio di Firenze in 1995. The web site was simple, it belonged to the first generation, presenting only the traditional

promotional brochure on a new support. Many other banks quickly followed the example of the Cassa di Risparmio di Firenze, by creating online pages and sites that proposed the benefits and advantages offered by the financial institution. By the end of 1995, 24 banks were present on the Internet. In 1996, more than 70 banks had their own online website, but they did not yet offer e-banking services (Carignani et al., 2007).

In 1998, according to data presented by IBM, over 84% of world banks offered e-banking services; this percentage rose to 92% in the 1999. It was only in 1998 that the first e-banking services began to appear: services mostly about consultation, not allowed to operate directly and to carry out transactions on accounts. The decisive year for the development and dissemination of e-banking services in Italy was 1999, the year of the great speculation of the new economy, the year of the online trading boom. It was in 1999 that many Italian banks became aware of the limitations of its online services and of the very high demand for advanced services.

9

countries that has always been ahead of his time in everything that could make life easier for citizens.

Three different phases could be identified through which the process of technological innovation has gradually approached banks and customers in order to provide the virtual access to an account from any location. First stage : access through a dedicated software. In the early 2000s, customers, in order to access their bank accounts, they had to use a specific link program provided free of charge by the bank. Through this software it was possible to issue orders (as, for example, carry out transfers or buy or

sell financial instruments) or receive information on their accounts in real time. The opportunity to interact with the bank through the use of a personal computer, represented a

great change from the past. This mechanism had two drawbacks, which would be resolved in the following stage. Firstly, if the customer had more accounts in different banks, he would necessarily use more programs to connect with each of them. Secondly, if the customer was using several PCs, he had to install the necessary software to connect on each of them. In sum, the customer could access his bank only through his PC, on which he had previously installed the required software. With the passage of time and user’s need to connect with his bank using different computers, banks have changed the means of access to their accounts.

Second stage: access through Web. The transfer of the IT platforms on Web-based applications has allowed access to their accounts from any PC connected to the network. To be able to access the

bank, it is sufficient that the reference station has an active connection to Internet. The access from any PC was finally possible but there was the possibility that the PC used was infected with

malicious software that could easily intercept the access codes. This led to the necessary

countermeasures prepared specifically by cyber bank technicians. While in the first phase to access accounts was sufficient to use an identification code and a

10

tablet in order to gain access to all transactions of the bank, from any place, for providing

instructions or simply to consult the balance account. In 2016 the last phase of the development of internet banking is faced, characterized by

multi-channel and multi-access services. In other words, the customers not only can access their accounts through different devices (PCs, tablets and smartphones), but they can use them in an alternative and complementary way.

2.6 Internet Banking in Romania

The period between the 1980s and the 1990s was marked by great changes in the territory of Eastern Europe. In fact, in those years Romania has eliminated the communist regime and this action had different consequences on the economic and politic system. The high inflation, the high level of unemployment and the consequent economic instability were the main features of that period (Buckley and Ghauri, 1994). Only in 1999 there were signs of recovery, with the increase of the GDP and the revaluation of the national currency. Regarding the Romanian banking system, between 1998 and 1999 seven commercial banks have been erased from the Bank register and several others have experienced financial problems. In 1998 banks moved from a state-owned system to a market system. This period was very difficult and the situation has improved only in 2000 and 2001. In fact, the development of private commercial banks was allowed by the introduction of a complex set of regulations regarding the role of the National Bank of Romania (Isarescu, 2001). Nowadays, the banks which are active in Romania can be divided into three categories. First, banks that have an extensive coverage of the territory (e.g. The Romanian Commercial Bank). Second, banks that are concentrated more in in a specific

geographical region (e.g. Carpatica). Third, foreign banks with branches in Romania (e.g. ING). Considering the Romanian banking industry, banks that have implemented Internet banking,

adopted a gradual strategy in order to introduce the new systems and services. First, they created an intra-organizational information network at national level. Second, they introduced credit and debit cards. Third, they introduced low-cost online banking services. Fourth, they introduced limited mobile banking services. Fifth, they instituted a connection between information systems at national level in order to create a good e-banking security (Gurau, 2002).

11

3. Methodology and Data

This section begins with the formulation of the research hypotheses. Later on, the data sample and time period, selection and collection will be discussed. In the final, the method of analysis will be discussed followed by an explanation of the dependent and dependent variables.

3.1 Hypotheses

In spite of the conspicuous use of the Internet as a delivery channel, few studies are available regarding an empirical analysis of the impact of the Internet on banks’ financial performance. More precisely, there are many studies in the early years of using Internet banking but they can no longer be considered valid in terms of estimation. The profitability factors may increase or decrease even after many years since the introduction of something innovative. Furthermore, Internet banking is offering new products even nowadays. Most of the research focuses on a single country. Few authors have tried to make a combination of two different countries in terms of banking culture. This paper is explaining that Romania had implemented internet banking during the sample period 2003-2013 and the effects of this implementation are analyzed. Moreover, Romania joined the European Union on 1 January 2007. The latter date is part of the sample taken in consideration. Romania being part of the European Union leads to the end of the segmentation of the financial market and the increase in stability and confidence in the banking system. In fact, joining the European Union means also to have a single supervisory mechanism and a single resolution mechanism (Paja, 2015). At the end of 1998, several banks state-owned were privatized and they became market-banks (i.e. Romanian Bank for Development and Bancpost). This is a substantial difference regarding the comparison with the Italian banks. In fact, state-owned banks are weaker than the market-banks and they have less competition. The low level of competition explains a low interest in satisfying the customers. In addition, the customers’ needs leads to the implementation of Internet banking.

This research attempts to fill this gap by identifying and analyzing the impact of the adoption of a transactional web site on financial performance using a sample of banks in Italy and Romania over the period 2003-2013. Banks operating in Italy share the same characteristics such as their universal character with western European banks. On opposite, results on banks operating in Romania could be extrapolated to the broader eastern banking system.

12 From the research question, three hypothesis arise: Hp1 : Internet Banking affects ROA positively.

The adoption of Internet banking has a positive relationship with the Return on Assets ratio. An increase in ROA means a higher profitability measure for the bank.

Hp2: Internet Banking affects ROE positively.

The adoption of Internet banking has a positive impact on the Return on Assets ratio. The increase in ROE leads to a more profitable measure of the bank.

HP3: Internet Banking affects Cost to Income Ratio negatively.

Internet banking has a negative association with the Cost to Income ratio. The decrease in the Cost to Income ratio means a higher measure of efficiency for the bank.

3.2 Data sample: time period, selection and collection

The study is focused from 2003 to 2013, considering a period of 10 years on the basis of 37 active commercial banks. The study considers two European countries as Italy and Romania. Data are collected from different sources. They were taken from Fitch’s database Bankscope Bureau Van Dijk for banks performance variables and for the cost-to-income ratio. Regarding the Internet-related details, this study will take in consideration information from banks’ websites and prior papers. The data concerning the implementation of Internet banking in Romania are extrapolated from the paper “The impact of internet banking on the performance of Romanian banks: DEA and PCA approach” (Stoica et al., 2013). On the other hand, the data regarding the implementation of e-banking in Italy were obtained through an accurate study on the specific web sites of the banks. Some banks do not share this type of data and therefore information were required by direct e-mail. The data’s study will be limited to only those banks for which explicit information on their Internet activities are available. Banks from the sample have to present consistent performance and asset-liability management variables during the entire 2003-2013 period. The final panel data comprises thirty-seven banks, nine Romanian banks and twenty-eight Italian banks, which represent over 90% of the banking asset of the Italian banking organization and 80% of the banking asset of the

13

banking activity. In addition, the majority of the Italian banks are native in that territory. In Romania, the majority of banks were created by foreign banks, in some cases even Italian (i.e. Unicredit Bank).

The analysis of the sample of banks is based on several financial performance ratios. These financial ratios measures profitability (return on assets ROA, return on equity ROE) and cost-to-income ratio. All the data regarding this ratios are taken from the database Bankscope for the period 2003-2013.

Regarding the coverage of Internet in Italy and Romania, Internet World Statistic reports some numbers. In 2000, in Italy, on a population of 57,989,000; 13,200,000 were internet users. In the same year, in Romania, on a population of 22,217,700; 800,000 were internet users. In 2010, Italy increases its internet users to 30,026,400, more than 51% of the population used Internet. In the same year, Romania reached 7,786,700 internet users, more than 35% of the population. 3.3 Variables

The dependent variables presented in the model are Return on Assets, Return on Equity and Cost to Income ratio. The measure of performance used in this research have been used in prior research. Hernando and Nieto (2006) explain the impact of Internet channel on banks’ profitability through the improvement of ROA and ROE.

ROE analyzes how much profit a bank earned compared to the total amount of shareholders’ equity. Return on equity measures a corporation’s profitability by revealing how much profit a company generates with the money shareholders have invested. ROE is expressed as a percentage and calculated as: Return on equity = Net income / Shareholders’ equity. Hall and Weiss (1967) argue that Return on Assets is the most widely used to lead at a consistent result that explains the

performance of a bank. Hall and Weiss (1967) claim that ROE do not differ among banks, that the

value it acquires is similar. This can lead to a better comparison in the sample. ROA is calculated by dividing a bank’s annual earnings by its total assets. Return on assets is an

14

number of factors can affect the ratio, including a bank's business model and size. Cost to Income ratio is expressed as a percentage and calculated as: Cost to income = Operating Costs / Operation Income. Arnaboldi et at. (2008) in their comparative analysis among Finland, Italy, Spain and UK

take into account the cost-to-income ratio as a measure of bank efficiency. The variable of interest for the model is the Internet adoption. The Internet is the dummy variable, it takes the value of 0 if the Internet banking is not adopted an the value of 1 if the Internet banking is adopted. Al-Sa’ Di and Khwarish (2001) use Internet as a matrix of dummy variables that equal 1 if the bank adopted Internet, otherwise 0 if the there is no trace of e-banking adoption. The control variables for the model are Loan to deposit Ratio and Equity to total Assets. Loan to deposit ratio calculates the bank’s liquidity by dividing the bank’s total loans by its total deposits. The higher the ratio, the lower the liquidity to fund requirements; the lower the ratio, the lower the bank’s earnings. Arnaboldi et al. (2008) use this measure as control variable in order to control for the relationship between funding and deposits. Equity to total assets is calculated by dividing total shareholder’s equity by total assets of the firm. Doltu (2000) argue thatthe equity to assets ratio is one of the financial ratios used to determine the financial health and long-term profitability of a bank.

3.4 Method

The model is based on the study of Al–Sa’ Di and Khwarish (2011) which was the continued research of Demirguc-Kunt and Huizinga (1999) and by Athanasoglou (2008), and by Aburime (2008). They define bank performance by measuring the Return on Assets and Return on Equity. The analysis contains different regressions in order to test whether the coefficient of Internet adoption is significant for the profitability measures and the cost-to-income ratio. The adoption of Internet is regressed over those performance variables and two control variables are introduced in the mode. The regression containing ROE, ROA, cost-to-income ratio, Internet adoption and the two control variables are run for the whole sample of banks.

The equation considered is the following:

15

COST TO INCOME RATIO = β0 + β1 INTERNETit + β2 LTDit + β3 E/TAit +

µ

+ λ

+ ε

The Yit is defined for bank performance i at time t. As profitability ratios, the study consider return on equity (ROE), return on assets (ROA), and cost to income ratio;

INTERNETit represents the dummy variable based on the time of e-banking adoption. This dummy variable equals 1 if the bank has a transactional website and 0 otherwise;

LTD is the control variable Loan/ Customer Deposits; E/TA is the control variable Equity/ Total Assets; β0 is the constant;

β1, β2 and β3 are the regression coefficients; µi the vector of bank fixed-effects;

λt the vector of period fixed-effects; εit the error term.

According to Hasan et al. (2008) period fixed-effects and bank fixed effects are included in order to prevent problem of multicollinearity. The period fixed-effects avoid distorting the regressions due to significant macroeconomic changes in the period under consideration. Bank fixed-effects control for banks’ characteristics in order to avoid major changes in the banks’ sample.

Considering the International dimension that this research underlines, the sample will be divided in two sub-samples. The first contains 9 Romanian banks, the second represents 28 Italian banks. The variables taken into account are the same and also the consequent regression. The purpose it to explain the institutional difference between the two countries, which creates mixed results in the findings. The hypotheses to be tested are those explained previously, the key variable is the

16

Descriptive statistics concerning the sample are given in Table 1. The total sample includes 37 banks, 9 banks located in Romania and 28 banks located in Italy. The panel data that has been created after collecting data from Bankscope result. Considering that the database was missing some observations, especially for the year 2006 regarding the Italian banks, data on the Cost to Income ratio and Return on Assets were taken also from the financial report presented on the webpage. Observations that were outliers were delated, following Hasan et al. (2008) calculations for mean and standard deviation. The cost to income ratio represents cost efficiency and the average in the sample is 67.24%, reflecting the findings of Arnaboldi in 2008.

TABLE 1 Descriptive Statistics

Mean Std. Deviation Median Minimum Maximum Return on Assets (%) 0.477 10.938 0.470 -4.350 3.980 Return on Equity (%) 5.489 10.973 0.477 -62.920 41.140 Cost to Income Ratio (%) 67.242 16.912 66.890 5.680 162.200 Loan/Customer Deposits (%) 137.935 51.709 144.350 9.180 502.440 Equity/Total Assets (%) 9.317 7.301 7.955 1.300 71.140 Internet 0.926 0.262 1.000 0 1.000 Observations 406 406 406 406 406

Notes: Total observations are 9 banks in Romania and 27 banks in Italy, considering the period between 2003 and 2013.

Table 2 reports the average of Return on Assets, Return on Equity and Cost to Income ratio per year. The Return on Assets is very small in 2003 but it increases in the following year. During the crises between 2007 and 2008 is slightly lower but it increase again until 2011. Return on Equity increased rapidly in the first years of the sample but it decreases after the crisis, leading to a very low level. Cost-to-income ratio has a similar pattern for almost every year. In Appendix A are presented graphs with the variation on a timeline of Return on Equity, Return on Assets and Cost to Income ratio. Graphs explain better the increase and the decrease of the variables taken into

17

TABLE 2

Evolution of ROA, ROE and Cost to Income Ratio

2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

Return on Assets (%) 0.45 0.92 0.81 0.84 0.93 0.77 1.99 2.10 2.14 0.02 -0.14 Return on Equity (%) 2.93 9.78 9.19 10.31 11.82 9.04 4.61 2.64 2.34 0.37 -3.21 Cost-to-income (%) 73.37 72.69 69.12 67.87 64.02 64.63 64.16 63.84 70.69 64.97 62.71 Notes: The average of the dependent variables between 2003 and 2013.

The following Table (3) shows the correlation between the variables considered in the model. The variable Cost to Income ratio is negatively correlated with the two profit variables that are Return on Assets and Return on Equity. Furthermore Return on Assets is highly correlated with the other measure of profitability, Return on Equity but this is not surprising because both are calculated taking in consideration the Net Income of the bank, according to Al-Sa’Dì and Khrawish (2011). The general correlation between the other variables considered in the sample is approximately low, mostly negative. The variable Loans/Customer Deposits is correlated negatively with the other variables, meaning that the variables are totally different from each other. The variable Equity/Total Assets has 0.075 of correlation with Return on Assets reflecting the number of Total Assets

presented in the two ratios.

TABLE 3 Correlation Matrix

(1) (2) (3) (4) (5) (6) Return on Assets (%) 1.000

Return on Equity (%) 0.895 1.000

Cost to Income Ratio -0.371 -0.366 1.000

Equity/Total Assets (%) 0.075 0.040 -0.444 1.000

Loan/Customer Deposits (%) -0.308 -0.244 -0.003 -0.099 1.000

18

4. Analysis and Discussion

The results obtained from the regression analysis test the hypothesis presented above and they are compared with the findings of the previous literature. First, results on the whole sample are shown, banks during the period 2003-2013 for Romanian and Italian banks. Second, in order to explain in a better way the comparative analysis, the sample is divided in two minor samples, one with the Italian banks and the other one with the Romanian banks.

4.1 Regression Analyses Results

Table 4 represents the regression analysis results of the cost to income ratio, return on assets and return on equity as dependent variable. Year and bank fixed effects are included in the regression analysis. Column 1 shows the testing of hypothesis 1, showing the impact of Internet banking on Return on Assets. Column 2 shows the results for hypothesis 2, testing the impact of e-banking on Return on Equity. Column 3 represents the results of testing hypothesis 3, testing of Internet

banking on Cost to Income ratio. The coefficient associated with the key variable, Internet, explains

the relationship between Internet adoption and bank performance. With the Return on Assets as dependent variable, a negative coefficient for Internet is presented and

there is no significance. With the Return on Equity as dependent variable, in the second regression, a positive coefficient for the dummy Internet adoption is analyzed but there is no significance regarding the p-value. Internet adoption has no impact on both the performance variables taken in consideration in the model. This not support Hasan et al. (2008) that show a positive significance of the dummy Internet adoption in the association with the performance variables, taking in

consideration a sample of Italian banks with a different period 1993-2000. Regarding cost-to-income ratio as dependent variable, the key variable Internet adoption results significant with a negative coefficient. A low level of cost-to-income ratio means a higher level of the bank’s efficiency. This is in line with the findings of Arnaboldi et al. (2008) that find a negative

impact of Internet adoption on cost-to-income ratio. Analyzing the results of the control variables, the ratio Loans/Deposits is significant in the three

regressions, for all the dependent variables and it has a negative coefficient for Return on Assets and Return on Equity. Moreover, the ratio Equity/Total Assets is significant only for the

profitability measure Return on Assets and for the Cost to Income ratio as dependent variable. To summarize, different results are presented after the regression analyses. On the one hand,

19

that find not consistent results in adopting internet banking on profitability measures. The sample contains Italian banks and the period analyzed is 1998-2000. In the end, Hypothesis 3 is not rejected, finding a negative effect between the dummy Internet and the variable cost-to-income ratio. After rejecting hypotheses 1 and 2, the positive relationship between Internet adoption and ROA and ROE described by Hernando and Nieto (2005) is not supported. This can be given by the fact that the sample is different and also the time period. In addition, Hasan et al. (2008) argued a positive link between Internet adoption and bank profitability and the sample contained Italian banks, but the time period was different. Moreover, Arnaboldi et al. (2008) with a comparative analysis between different countries, including Italy found no correlation between using Internet banking and having a high performance indicator.

TABLE 4

Coefficient Estimates from Regression Analyses Results

Dependent Variable Return on Assets Return on Equity Cost to income ratio Constant 1.348*** 7.778** 67.648*** (0.293) (3.522) (4.562) Internet -0.142 1.102 -6.787** (0.182) (2.183) (2.827) Loans/Deposits -0.006*** -0.030* 0.077*** (0.001) (0.017) (0.022) Equity/Total Assets 0.013* 0.084 -0.520*** (0.009) (0.112) (0.145) Year Fixed Effects Yes Yes Yes Bank-Fixed Effects Yes Yes Yes R2 _{0.494 0.467 0.623} Adjusted R2 _{0.422 0.392 0.571} F-statistic 6.932 6.218 11.787 P-value 0.000 0.000 0.000 N 406 406 406 Notes: Column 1 shows the testing of hypothesis 1, testing the impact of Internet on Return on Assets; Column 2 shows the results for hypothesis 2, testing the impact of Internet on Return on Equity;

Column 3 represents the results of testing hypothesis 3, testing the effect of Internet on Cost to Income ratio; *** significant at 1 percent; ** significant at 5 percent;

20

In order to test the three hypotheses with separate samples, more regression are run. Table 5 shows the regression analysis results for the variables ROE, ROA and cost-to-income ratio. Year and bank fixed effects are included in the regression analysis. Column 1, 2 and 3 show the results for

hypothesis 1, 2 and 3, respectively regarding the sample that contains the Italian banks. Column 4, 5 and 6 show the results for hypothesis 1, 2 and 3, respectively in relation with the sample that

presents the Romanian banks. The key variable Internet adoption acquires significance only with the cost-to-income ratio as dependent variable in the sample with the nine Romanian banks. In fact, this reflects the result obtained previously in the sample with 406 observations. In addition, the coefficient for Internet is negative and is higher than the one found in the sample with 406 observations (-8.878 > -6.787). In sum, Hypothesis 3 is not rejected for the regression analysis regarding the secondo sub-sample, Romanian banks. On the contrary, Hypothesis 3 is rejected for the first sample, the coefficient of Internet is not significant at any critical level. Concerning the two measures of profitability, ROE and ROA are not affected by the adoption of Internet, the coefficient is not significant for both the countries. Hypothesis 1 and 2 are rejected for the two sub-samples. The control variable Loan/ Deposits is significant in the Italian sample for every dependent

variables, it assumes a negative relationship with the two profitability measures and a positive with the cost-to-income ratio. This is in line with the findings of Arnaboldi et al. (2008). The other control variable, Equity/Total Assets is significant in the Italian sample with cost-to-income ratio as dependent variable and in the Romanian sample with ROA as dependent variable. In both cases, it assumes a negative relationship.

TABLE 5

Coefficient Estimates from Regression Analyses Results

ITALY ROMANIA

21 R2 _{0.485 0.457 0.614 0.594 0.626 0.719} Adjusted R2 _{0.405 0.373 0.554 0.483 0.524 0.642} F-statistic 6.084 5.440 10.309 5.362 6.129 9.374 P-value 0.000 0.000 0.000 0.000 0.000 0.000 N 307 307 307 99 99 99 Notes:

Column 1 shows the testing of hypothesis 1, testing the impact of Internet on Return on Assets on Italian banks; Column 2 shows the results for hypothesis 2, testing the impact of Internet on Return on Equity on Italian banks; Column 3 represents the results of testing hypothesis 3, testing the effect of Internet on Cost to Income ratio on Italian

banks;

Column 4 shows the testing of hypothesis 1, testing the impact of Internet on Return on Assets on Romanian banks; Column 2 shows the results for hypothesis 2, testing the impact of Internet on Return on Equity on Romanian banks; Column 3 represents the results of testing hypothesis 3, testing the effect of Internet on Cost to Income ratio on

Romanian banks; *** significant at 1 percent;

** significant at 5 percent; * significant at 1 percent;

5. CONCLUSION and LIMITATIONS

22

banks and one for the twenty-seven Italian banks. The adoption of Internet is not significant in relation with ROE and ROA in any of the two subsamples. Regarding the cost-to-income ratio Internet presents a significant coefficient, being negative it increase the efficiency of banks. This is in line with the research made by Arnaboldi et al. (2008). Conversely, the sub-sample for the Italian banks reveals no significant results in the regression on the relationship between Internet and

performance ratio and on the association between Internet and cost-to-income ratio. This means that the significance of Internet adoption on cost-to-income ratio is given by the introduction of the Romanian banks in the general sample. According to Al-Sa’Dì and Khrawish (2011) the regression analysis reveals no significant impact of adopting Internet banking on the banks’ performance. 5.1 Limitations

There are several limitations for this research. First, the selection bias that may be present in the research is the sample. In fact, in Italy banks are completely different from each other and they have different performance results. Future research may investigate another sample, with more banks or banks not so different from each other in size and types. A clear classification of banks may lead to more reliable results. Second, the methodological bias that may be present in the paper is the

selection of the control variables. In fact, prior literature focused on more control variables that can affect the dependent variables. Future research may consider control variables that are more related to the adoption of Internet. Furthermore, this paper has not focused on the crisis years that are also present in the sample between 2003 and 2013. Indeed, the effects of the crisis might have had an impact on profitability banks.

5.2 Recommendations for management

23 6. REFERENCES

Aburime, T., 2008. Determinants Bank Profitability: Macroeconomic Evidence from Nigeria. Deakin University, 1-10. Available at: https://www.researchgate.net/publication/228282854.

Athanasoglou, P., P., Delis, M., D., and Staikouras, C., K., 2008. Determinants of Bank Profitability in South Eastern European Region. Available at http://mpra.ub.uni-muenchen.de/10274.

Al-Sa’dì, N., M., Khrawish, H., A, 2001. The impact of e-banking on bank profitability: Evidence from Jordan. Middle Eastern Finance and Economics

Bonaccorsi di Patti, E., Gobbi, G., Ministrulli, E., P., 2004. Testing for Complementarity between Stores and E-Commerce: The Case of the Banking Industry. Banca d’Italia.

Buckley, P.J., and Ghauri, P.N., 1994. Statement of the issues. The Economics of change in East and Central Europe, Academic Press.

Carlson, J., Furst, K., Lang, W., Nolle, D., 2000. Internet Banking: Markets Developments and Regulatory Issues, Economic and Policy Analysis Working Papers, 2000-9, Office of the Comptroller of the Currency.

Carignani, A., Gemmo, V., 2007. Prestiti peer to peer: Modelli di business e strategie. Credito Popolare, 14(3-4), 409-425.

Claeys P., Arnaboldi, F., 2008. Internet Banking In Europe: a comparative analysis. Available at: https://ssrn.com/abstract=1343684.

Cyree, K., Delcoure, N., Dickens, R., 2009. An examination of the performance and prospects for the future of internet-primary banks. Journal of Economics, and Finance, 33(2), 128-147.

Dandapani, K., G. Karels, V., Lawrence, E.,R., 2008. Internet banking services and credit union performance, Managerial Finance, vol. 34, no. 6, 437-446.

Delgado, J., Hernando, I., Nieto, M.J., 2006. Do European Primarily Internet Banks Show Scale and Experience Efficiencies. European Financial Management.

Demerguç-Kunt, A., Huizinga, H., 1999. Determinants of Commercial Bank Interest Margins and Profitability: Some International Evidence. World Bank Economic Review, 13, 379-408.

24

Doltu, C., 2000 .The evolution of the banking system in Romania. Romania 2000-10 years oftransition: past, present and future.

Egland, K.L., Furst, K., Nolle, D., Robertson, D., 1998. Banking over the Internet. Office of the Comptroller of the Currency Quarterly Journal 17 (4), 25–30.

Furst, K., Lang, W.,W., Nolle, D., E., 2002. Internet banking. Journal of Financial Services Research 22, 95–117.

Gurau, C., 2002. E-banking in transition economies: the case of Romania. Journal of Financial Services Marketing 6 (4), 632-378.

Hall, M., Weiss, L., 1967). Firm Size and Profitability. The Review of Economics and Statistics, 49 (3), 319 – 331.

Hasan, I., Zazzara, C., Ciciretti, R., 2008. Do Internet Activities Add Value? Evidence from the Banking Industry. Rensselaer Polytechnic Institute.

Hernando, I., Nieto, M., J., 2005. Is the Internet Delivery Channel Changing Banks’ Performance? The Case of Spanish Banks. Banco de Espana, unpublished manuscript.

Hoffman, D.L., Novak, T.P. and Peralta, M., 1999, Building customer trust online, Communication of the ACM, Vol. 42 No. 4, 80-85.

Internet World Stat. Retrieved at : http://www.internetworldstats.com/eu/ro.htm. Accessed on 14-11-2016.

Isarescu, M., 2001. Dezvoltarea activitatti de e-banking este, fara indoiala, un element important in evolutia sistemului financiar romanesc. E-Finance Romania.

Lambrecht, A., Seim, K., 2006. Adoption and Usage of Online Services in the Presence of Complementary Offline Services: Retail Banking. Available at:

https://archive.nyu.edu/handle/2451/28475.

Lu, M., Liu, C., Jing, J., Huang, L., 2005. Internet banking: strategic responses to the accession of WTO by Chinese banks. Industrial Management and data system 105(4), 429-442.

KPMG, 2016. The need for speed, 2016 Banking industry outlook survey. Retrieved from :

25

Mattila, M., Karjaluoto H., Pento, T., 2003. Internet banking adoption among mature customers: early majority or laggards? Journal of Services Marketing, Vol. 17 No. 5, 514-528.

Onay, Ceylan. Ozsoz, Emrr. Helvacioglu, Ash Debiz, 2008. The impact of internet banking on banks profitability. The case of Turkey. Oxford Business and Economics Program.

Sahut, J., Kucerova, Z., 2009. Enhanced internet banking service quality with quality function deployment approach. Journal of Internet Banking and Commerce.

Sathye, M., 2005. The impact of internet banking on performance and risk profile: Evidence from Australian credit unions. Journal of Banking Regulation 6, 163–174.

Sullivan, R., J., 2000. How has the adoption of Internet banking affected performance and risk at banks? A look at Internet banking in the tenth federal reserve district, Federal reserve bank of Kansas City. Financial Industry Perspectives, 1–16.

Tsai, C., H., Zhu, D., S., Jang, Y., M., 2013. A study on the consumer adoption behaviors of Internet bank. Journal of Networks, Vol. 8 No.12, 2956-2963.

26 Appendix A

Sample

TABLE 6

Bank’s Country of Residence BANKS

ITALY ROMANIA

Carige Alpha Cassa di Risparmio di Firenze Carpatica Dell’Adriatico BancPost Del Piemonte Transilvania Del Fucino BCR Di Bologna CEC Generali Piraeus Monte dei Paschi di Siena ProCredit Monte Parma Raiffeisen Nazionale del Lavoro

Nuova

27 Appendix B

Evolution of ROA, ROE and Cost-to-Income Ratio

GRAPH 1 Return on Assets

Notes: Return on Assets has a rapid increase after 2008 but a dramatic decrease in the last year of the sample.

GRAPH 2 Return on Equity

28 GRAPH 3 Cost to Income Ratio

56 58 60 62 64 66 68 70 72 74 76 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Rat io (% ) Year