IMPACT of FINANCIAL INNOVATION on the FINANCIAL PERFORMANCE of the TRADITIONAL FINANCIAL INTERMEDIARIES a Thesis Presented To

Running head: FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES

IMPACT OF FINANCIAL INNOVATION ON THE FINANCIAL PERFORMANCE OF THE

TRADITIONAL FINANCIAL INTERMEDIARIES

A Thesis

Presented to the Faculty

of ISM University of Management and Economics

in Partial Fulfilment of the Requirements for the Degree of

Master of Financial Economics

by

Luka Juodelytė

May 2018

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 2 Juodelytė, L., Impact of Financial Innovation on the Financial Performance of the Traditional Financial Intermediaries. [Manuscript]: Master thesis: Financial Economics. Vilnius, ISM University of Management and Economics, 2018. Abstract The purpose of this thesis is to examine the impact of digital bank deposit, asset and loan growth on selected traditional bank performance measures. Review of previous literature allows to verify an existing gap in the literature for such research. Analysis of previously performed research on this topic assists in determining traditional bank performance measures which can be affected by changes in digital bank factors – return on assets, return on equity, net interest margin, net non-interest margin, loans to assets and liabilities to equity. In order to estimate whether a causal relationship between digital bank measures and traditional bank performance exists, Granger causality method is selected as the main empirical model. In addition, to determine the direction and strength of said relationship, OLS regressions are performed.

Research results lead to the conclusion that digital bank deposit and loan growth have a causal relationship to traditional bank performance ratios. Deposit growth has a negative impact on traditional bank performance ratios and loan growth shows both positive and negative impact on different ratios. This research demonstrates some of the challenges that traditional banks are facing in the age of innovation. As deposit and loan growth are perceived as proxies to customer growth, focus areas identified for traditional banks to safeguard profit and market share are customer attraction and preservation. As the research has some limitations, such as excluded regulatory effects and using growth data instead of level data, future research is recommended to include regulatory changes and test level data instead or in addition to growth data. (20 487).

Keywords: financial innovation, digital banks, traditional financial intermediaries,

Granger causality. FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 3 Table of Contents

List of tables ...... 5

List of figures ...... 6

Glossary ...... 7

Introduction ...... 8

Literature review ...... 13

Concept of innovation ...... 14

Innovation and economic performance ...... 16

Innovation and the financial sector ...... 17

Concept of financial innovation ...... 18

Financial innovation research review ...... 21

New products ...... 21

New processes ...... 27

New organizational form ...... 31

Summary of literature review ...... 36

Research methodology ...... 39

Performance measures ...... 39

Digital bank measures ...... 44

Research instrument selection...... 47

Granger causality ...... 47 FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 4 Ordinary least squares regression ...... 50

Data selection ...... 51

Methods for data testing...... 54

Possible limitations of the research ...... 56

Empirical research results and discussion...... 60

Descriptive statistics ...... 60

Normality tests ...... 61

Granger causality ...... 64

Ordinary least squares regression analysis ...... 66

Discussion ...... 69

Conclusions ...... 77

References ...... 80

Appendices ...... 94

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 5 List of tables

Table 1. Performance measures of banking in literature...... 41

Table 2. Econometric hypotheses ...... 48

Table 3. Traditional banks by asset size ...... 52

Table 4. Digital banks by asset size ...... 53

Table 5. Summary statistics of the un-adjusted data ...... 60

Table 6. Summary of Granger test results...... 64

Table 7. Summarized results of OLS analysis ...... 67

Table 8. Summary of research results ...... 69

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 6 List of figures

Figure 1. Noninterest and net income as a % of total operating income in U.S. commercial banking, 1970–2003...... 44

Figure 2. Correlation matrix of exogenous variables ...... 63

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 7 Glossary

ROA Return on Assets

ROE Return on Equity

NIM Net Interest Margin

NNIM Net Non-Interest Margin

ECB European Central Bank

ATM Automated Teller Machine

ACH Automated Clearinghouse

OECD Organization for Economic Cooperation and Development

NACHA National Automated Clearinghouse Association

ADF Augmented Dickey-Fuller test

KPSS Kwiatkowski–Phillips–Schmidt–Shin test

OLS Ordinary least squares model

NFA New Financial Architecture

MBS Mortgage Backed Security

ABS Asset Backed Security

AIC Akaike Information Criterion

BIC Schwarz-Bayesian Information Criterion

HQC Hannan-Quinn Information Criterion

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 8 Introduction

We are living in times of innovation and development and witnessing constant change marked by the emergence of new business models in various industries. Innovation is changing the face of business and improving its conditions by creating value for the shareholders. There’s no arguing that innovation is one of the most important variables in economic growth. Solow

(1956) states that economic development, in the long run, is impossible without innovation and innovation is the only variable which impacts growth. Even though Solow discusses technological innovation, rather than financial, Blach (2011) argues that without financial innovation technological and economic development would be much slower and as a consequence, wealth of nations would be lower. Thus, financial innovation is critical, both, for the business sector to spur development and growth, which increases shareholder value, and for the public sector to increase the economic growth and thus, the standard of living.

According to Aspara, Rajala & Tuunainen (2012), the financial industry has been immensely transformed by changes in regulation, globalization and digitalization. These changes have highly increased the attention to customer experience and moved the financial industry from using technology only as facilitator of internal processes to being a crucial component in value offering. Hence, in today’s world where technological innovation can hardly be separated from other forms of innovation, including financial, it becomes even more important to understand how it impacts the more regulated and thus, less prone to change, traditional industries like banking.

The banking sector has always been one of the most vital organs of the economy. Both, the private business sector and the public sector relies on banks for loans and investment opportunities. However, banks not only take deposits and create credits but are also a key FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 9 element to keeping a stable economy in the region. Initially, economic stability is controlled by

Central Banks (EU) and Federal reserve (US) through contractionary monetary policy, which manages money supply through bond prices and interest rates. However, Central banks and

Federal reserve implement this monetary policy through commercial banks, thus, commercial banks play a key role not only for the private and public sectors, but in the overall economic stabilization process.

While banks are one of the most traditional businesses, they are also clearly in the center of the financial world. This makes them more exposed to changes in the industry and the whole economy. According to Shirakawa (2011), innovation in the financial industry has been an ongoing process for centuries – beginning with first stock exchanges and continuing with more recent innovations such as bank teller replacement by ATM’s and plastic cards, which are becoming increasingly more important than cash. The banking sector has always adapted and even benefited from innovation. However, nowadays new challenges are posed in front of traditional financial intermediaries as the changes in the industry are as rapid as never before with new competitors rising from the FinTech industry (“companies that use technology for banking, payments, financial data analytics, capital markets and personal financial management”

(Huang, 2015)), which has been growing as fast as 201% in 2014 (Dickerson, Skan&Masood,

2015).

Growth of financial innovation in inevitable and, moreover, highly needed for economic development. But financial innovation, such as the emergence of digital banks supported by the growth of FinTech industry might threaten to take over the role of main provider of financial services from the traditional banks (Dabrowski, 2017). In 2015, 33% of customers were likely to switch banks in the upcoming year (PWC consumer banking survey, 2015). This leads to the FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 10 assumption that customers do not see switching banks as an obstacle. 11% of customers switched to internet only banks in 2015 (North America Consumer Digital Banking Survey, 2016), which indicates that they are, as expected, a viable competitor of traditional banks. In order to safeguard their place in the financial market, it is crucial for traditional banks to be aware of how innovation shapes the market and more importantly, how it impacts their performance. Thus, the impact of financial innovation on traditional financial intermediaries is significant for the private sector, the publics sector and the financial intermediaries themselves.

However, even though financial innovation is clearly an important field, there is lack of research made in this area. Most research made in the innovation field is focused on innovation in manufacturing industries. Another tendency seen in more recent research is considering financial innovation as technological innovation in the financial sector (FinTech). This direction is an improvement in the current situation of financial innovation research, however, it still makes up a small part of innovation research and rarely separates different types of FinTech

(such as digital banks, electronic money, cryptocurrencies) but handles it as one bundle. Thus, distinguishing digital banks as one of the types of financial innovation and performing an empirical analysis of its impact on traditional bank performance is a great improvement in the current situation of financial innovation research.

Therefore, the research question is: what is the impact of digital banks on the performance of the traditional banks? In order to answer the research question, the research goal of this study is set: analyze the impact of financial innovation, such as digital banks, on the performance of traditional financial intermediaries, such as traditional banks by concentrating on financial digitalization. This goal is achieved through the following objectives: FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 11 1. By analyzing the existing literature on financial innovation, define the

concept of financial innovation and create a theoretical background on the subject in

order to identify the gap in the current situation;

2. Based on previous research, identify the performance measures of

traditional financial intermediaries, which can be impacted by certain factors in

digital banks;

3. Identify an appropriate econometrical model to test digital bank impact on

traditional financial intermediaries, collect relevant data and determine the possible

limitations of the selected model and dataset;

4. Based on the defined research methodology, perform an empirical research

to verify whether digital banks challenge the performance of traditional financial

intermediaries’;

The thesis is designed as follows:

The literature on the topic of financial innovation is analyzed. In the literature review part, first, the concept of innovation is defined, its importance to overall economic performance is discussed and in this light, the concept of financial innovation is defined and types of it are classified. The most significant literature on the main innovations that are relatable to this thesis in each class of financial innovation (new products, new processes, new organizational forms) are discussed, identifying the gaps in the field.

After the literature review, the research methodology applied in order to reach the goal of the thesis, is discussed. In order to answer the research question, first, measures of performance studied in previous literature are analyzed and estimations are made which of them, in theory, could be impacted by digital banks. Then, economical hypotheses are presented. Further on, an FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 12 appropriate econometrical model is selected. Then, the scope, scale, sources of data and methods for data testing and modification are discussed. In the end of the research methodology chapter, the limitations and weaknesses of the methodological design are presented.

Finally, an empirical research is conducted in accordance with the methodological design discussed in the research methodology part and the results are presented. Moving forward, the results are discussed in the light of existing literature and the implications on existing theory and practical implications are presented as well as suggestions for further research and developments in the field are provided.

This research is concentrated on US market during the period of 2005-2017. US market is selected because digital banking in the US is more established than in Europe. For this reason, more extensive data can be found and the uncertainty of start-up banks surviving can be avoided.

The time period of 2005-2017 is selected partly to have a broad data set and partly due to the lack of empirical research after the financial crisis of 2007-2009. Traditional and digital banks are selected according to their size by assets and other data restrictions discussed in the research methodology part. The data is collected from the Bloomberg terminal.

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 13 Literature review

This part of the thesis is a review of academic literature that is relevant to financial innovation and traditional financial intermediaries field. The aim of literature review part is to familiarize the reader with the theoretical background on the concept of innovation, its impact on economic performance, the concept of financial innovation, its various categories and its impact on traditional banking as well as the critique by various authors on the topic and to form a strong base for designing a suitable methodological approach for further empirical research.

Financial innovation field does not have a long history of research and literature behind it - the field of financial innovation is quite recent in scientific terms as the beginning of research on it commences around 1970s. However, despite its novelty due to the rapid rate of innovation in the financial sector and the growing importance of the financial sector itself to the economy, there is a high interest in this field. An unexpected attribute of the field, however, is that although there are some studies conducted, few of them are empirical. Which is even more surprising given the quantity of such research in other fields associated with innovations such as manufacturing. Therefore, there is a gap for empirical research in the financial innovations field.

In order to provide a full insight on the topic, first, the concept of innovation in the general sense is discussed as well as its impact on overall economic performance. Then, the concept of financial innovation is considered. The most important financial innovation research is reviewed by splitting the innovation to three major categories (product, process, organizational form).

Although commercial banks, investment banks, mutual and pension funds are all classified as traditional financial intermediaries, this thesis concentrates on the impact of FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 14 financial innovation on the financial performance of commercial banks, thus, the literature review part is limited to commercial banks only.

Concept of innovation

To have a strong theoretical base on the subject of financial innovation and to be able to move further on with the research of digital bank impact on traditional banks, first it is important to understand the concept of innovation. Throughout the years, innovation has been assigned different definitions by authors in various fields of research.

Schumpeter (1934) is considered to be the first one to define innovation in scientific research. He defined innovation as a change in production system in order to increase profit and decrease costs. Acs & Audretsch (1988) defines innovation as “<…> a process that begins with an invention, proceeds with the development of the inventions, and results in the introduction of a new product, process or service to the market-place.”. Damanpour (1992) states that innovation is simply an adoption of an idea or a behavior. Hage (1999) agrees with Damanpour by defining innovation as “the adoption of an idea of behavior that is new to the organization” which can be a new product, service, technology or process. Fruhling & Siau (2007) have the same idea of innovation being a new idea, practice, behavior or object that is new to the perceiver of it. Geiger and Cashen (2002) narrow the definition down by stating that innovation is simply creation of a new product. Palmberg (2004) is even more specific by defining innovation as “a technologically new or significantly enhanced product compared to the firm’s previous product”.

Although Dibrell, Craig & Davis (2008) state that innovation cannot be restricted as it can vary in its complexity and scope from minor changes to existing products, processes or services to completely new products, processes or services. To summarize, to one extent or another, all the FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 15 definitions above state that innovation is a new invention or an improvement for an older invention that brings better value to the company, consumer and/or economy.

Innovations are classified in various ways. Most authors discussed in the paragraph above classify innovation to product, process and service. It can also be classified by its impact on the market (Chandy & Tellis, 1998) – incremental, market breakthrough, technological breakthrough and radical innovation. But the most widely used classifications for innovation is the one by

Joseph Schumpeter (1983). He has divided innovation into five major categories:

1. New product

2. New methods of production

3. New sources of supply

4. New markets

5. New ways of business organization

OECD (1997) has based its methodology on Schumpeter’s classification. OECD classifies innovation to four categories: product (1), process (2), marketing (3) and business organization (4). According to Anderloni & Bongini (2009), development in these categories is seen as innovation if it is new to the organization or economy implementing it. This means, that nevertheless whether a product or process has been already implemented in other organizations or economies, it still is perceived as innovation for the organization or economy which is implementing it. Considering the definitions presented in the previous literature, in this thesis innovation is defined as a new or significantly improved product, process or business organization which is new to the implementing body and brings better value to the customer and the company/economy implementing it. FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 16 Innovation and economic performance

Innovation, of any kind, is a crucial factor for the economy. For a long time, economic growth has been seen as a consequence of change in static factors. Although various authors agree that economic development in the long-run is driven by dynamic factors, such as innovation. Schumpeter (1934) presented the theory of Creative destruction which states that innovation, created by the entrepreneur, spurs economic growth. Solow (1956) was the first one to design a model that justifies innovations importance to the economic development. In his model Solow says that innovation is the only driver that can spur development in the long run.

Even though Solow discusses technological innovation, rather than financial, Blach (2011) argues that without financial innovation, technological and economic development would be slower. Scherer & Ross (1990) agree that no matter how important is static resource allocation in the short-run, in the long-run, dynamic factors, such as innovation, drive the economy.

The idea of evolutionary economics also contributes to the justification of the importance of innovation to economic growth. Nelson & Winter (1982) were the pioneers of innovation in evolutionary economics. In their model, firms search for new and undiscovered ways (1) or ways already used in other firms but not in their own processes (2) that increase profitability when firms rate of return falls below a certain value. As the importance of innovation to the overall economy has been studied by various authors to this day, it can be asserted that innovation and economic performance are indeed highly related.

Innovation and other sectors

In order to discuss the impact of financial innovation on traditional financial intermediaries, it is important to note, how non-financial innovation shapes the existing, traditional companies. Chandy & Tellis (1998) finds that companies which tend to focus on their FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 17 current customers and competition rather than future customers and competition, are more prone to incur loses or even seize to exist when radical innovation manifests. Therefore, it can be believed that the same theory is valid for traditional financial intermediaries. Based on consumer theory, Frank (2015) states that innovation can either complement old services, if used together, or replace them by meeting the same needs of the consumer better. Consumer theory combined with the research conducted by Chandy & Tellis (1998) leads to believe that the impact of financial innovation on traditional financial intermediaries is as well headed the same direction.

If the innovation can better serve future customers, it will raise a threat for traditional bank existence, however, if the innovation is jointly utilized with traditional financial intermediaries, it can serve as complementary to them.

Innovation and the financial sector

The financial sector is the center of the economy (Goldsmith, 1969; McKinnon, 1973;

Levine, 1997) and is key to economic growth. Acemoglu & Zilibotti (1997) state that financial depth helps mobilizing and pooling savings, which stimulates capital accumulation and assists in its allocation. Greenwood & Smith (1997) presented a model that shows how financial markets lead the economy to higher productivity and higher growth by reducing transaction costs, promoting specialization and reducing liquidity risk. On the other hand, demand-following hypothesis (Ireland, 1994) states an opposite opinion that a growing economy stimulates financial innovation and the growth of the financial sector. According to the hypothesis, when the overall economy grows, market participants require better and more developed financial services which fosters financial sector development. However, Greenwood & Smith (1997) argue that the causality between financial sector and economy might be mutual – growth in one causes growth in another and vice versa. FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 18 Innovation has been a relevant research topic over the years and it still manages to raise questions for academia, one of the most important being – how does innovation impact economic development? Financial sector and its relationship with overall economic performance in macro and micro environments has also received attention over the years. As discussed before, financial sector and economic growth are believed to have a causal relationship, whether it would be mutual or one-way. As a consequence, it would seem that it is only natural that financial innovation would be a critical research topic. But although both, financial sector and innovation, have been studied and there is a firm academic background on these topics separately as well as their correlation with economic growth and development, financial innovation has not received the proper attention it requires. In the following sections, the concept of financial innovation is presented and the scarce theoretical background on financial innovation is reviewed in order to emphasize the need of such research.

Concept of financial innovation

Mishkin (1990) discusses that financial innovation is no different than innovation in any other field as it is driven by the same factors as other types of innovation, precisely – changes in technology and market conditions. Although as the financial sector is highly regulated, financial innovation has one additional factor to it – regulation. However, financial innovation regulation is not discussed in this thesis as the topic of regulation itself is very wide and thus is out of scope for this particular research. Other sources, such as Simsek (2013), choose a very abstract definition – financial innovation is defined simply as the creation of new assets and reduction in transaction costs. In this thesis, a broader approach to financial innovation than the one by

Simsek is exploited and financial innovation is perceived as the new creation of a product, FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 19 process or organizational form (or a significant improvement of an already existing), which brings value to the company using it.

According to Cecchetti & Schoenholtz (2015), the main functions of financial intermediaries are pooling savings (e.g. accepting resources through savings accounts from savers and lending to borrowers), safekeeping and accounting (e.g. keeping the savings accounts safe, giving access to savers to payments systems), providing liquidity (e.g. allowing depositors to transfer their financial assets into cash quickly), diversifying risk (e.g. creating different investment opportunities in order to diversify risk), collecting and processing information (e.g. providing financial information to customers). Providing these functions involve various costs incurred by financial intermediaries and financial facilitators. Uncertainties about future due to numerous time periods involved also raise the risk which transfers to costs for financial system participants. As there are costs and risks involved, the demand for better designed products and processes that are likely to reduce costs and risks is always growing. Thus, financial innovation can be explained as new product or process that reduces costs, risks and/or better satisfies the demands of financial sector participants (Frame & White, 2004).

Horne (1985) studies the reasons for financial innovation. He affirms that financial innovation arises when profit opportunities are seen, but distinguishes 6 stimuli for financial innovation: volatile inflation and interest rates (e.g. demand for more risk-reducing financial products grows) (1); regulatory changes (e.g. removal of regulation encourages new processes/products) (2); tax changes (e.g. changes in interest and dividend taxation prompts to look for new ways of tax evasion) (3); technological advances (e.g. technological advances gives the possibility of more speedier, accurate and less costly processes) (4); level of economic activity (e.g. in prosperous times, more new ideas are tried out when in recessions, the focus FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 20 shifts to liquidity and risk reduction) (5); academic research on market efficiency (e.g. more trained professionals enter the market with new ideas) (6). Both Miller (1986) and Merton

(1992) in their research also agree that financial innovation is mostly driven by regulatory and tax changes.

In regard to previously discussed literature on innovation and financial innovation, in this thesis, financial innovation is defined as new or significantly improved product, process or organizational form developed for/by the financial sector, which is new to the implementing body (firm/economy) and brings better value to it.

In the previous section, the most widely used innovation classification by Schumpeter

(1983) has been provided. A simplified classification is adapted to financial innovation in this thesis:

1. New products (e.g. various types of securities, debit cards)

2. New process (e.g. automated clearing houses, internet banking)

3. New organization forms (e.g. internet-only banks)

The classification of financial innovation could be expanded, but for the sake of simplicity, a narrower classification is used in this thesis. The principal goal of this classification is to focus on the main research areas of financial innovation that are associated with financial intermediaries without going deeper into other areas such as marketing or business operations.

Various other forms of innovation could also be discussed – innovations, used in the financial sector, but primarily developed for/by a different sector (e.g. mobile application payments), however in this thesis they are out of scope. Only purely financial innovations - innovations developed for/by the financial sector - are discussed. FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 21 Financial innovation research review

The previously discussed importance of financial sector to the overall economy and the importance of innovation to economic growth raises the topic of financial innovation to be one of the most important topics of the 20th - 21st century research in finance. The research review section of the thesis is divided according to the innovation classification provided in the previous section – new products, new process and new forms of business organization.

New products

Financial innovation is often perceived and researched as security innovation not only in the 20th century, but to this day. Security innovation is the biggest, most researched product innovation in finance.

Even though, only innovations used by the financial sector are in scope of this thesis, it is worth shortly mentioning other perspectives on financial product innovation for a broader sense of the reader. One of such is an approach to product innovation taken by Faulhaber & Baumol

(1988). The authors have studied economists as innovators and the innovation made by them which improves the activities of non-financial sectors. Faulhaber & Baumol review financial theory models which were applied in practice as products in the non-financial sectors. Examples of this are Black-Scholes pricing model used by investors or Ramsey pricing model used by regulators. However, this approach focuses on non-financial sector, thus, it is not further discussed in the thesis.

Berger (2003) has made a huge impact on product innovation literature with his extensive descriptive research on innovation effect on the traditional intermediaries’ industry. Berger reviews the researches done on technological innovations used in the banking sector and how they have impacted banking sector performance. He finds that technological progress, such as FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 22 ATM’s and internet banking, which are considered as product (ATM) and process (internet banking) innovations, are proved to be positively linked with bank productivity and performance.

It is clear that bank performance is highly linked with new product innovation. Berger also finds that progress in the IT sector has increased the productivity and economies of scale together with reducing costs for electronic payments. This explains the shift in the banking industry from hard cash to digital money and the increase in usage of debit and credit cards by customers, which also can be interpreted as one of the main reasons for the emergence and growth of banking digitalization.

The product innovation discussed in the literature is mostly innovation implemented by banks and thus, impacts bank performance through its use by the intermediary itself. However, there is little to no literature on the innovation impact of banks performance when it is not used by the bank or is not a product sold in a bank or through a bank. The closest that product innovation literature comes innovations that can impact banks externally are pre-paid cards. In this section, literature on securities, debit cards and prepaid cards is reviewed.

Securities. Security literature can be divided to descriptive and empirical literature. Some descriptive researches, as the ones by Miller (1986; 1992) and Mishkin (1990), discuss the causality of security innovation, different types of innovation through time and possible faults of security innovation. Duffie & Rahi (1994) provide a great theoretical introduction to security design. In their article, authors give a wide overview of previous literature on security design.

They also provide a list of economic events and financial innovations in time to demonstrate the cause and consequence of security innovations. The authors not only discuss the newest financial products and how they are modeled but present the most used models for security research and their main principles, although they do not create any new models themselves. FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 23 However, most of the literature in the security design segment is empirical. It has one of the broadest range of empirical research in financial innovation literature, thus, it is worth mentioning. The empirical literature on security design mainly focuses on a few topics: risk sharing, reducing asymmetrical information issues (Boot & Thakor, 1993) and increasing liquidity for investors (DeMarzo & Duffie, 1999).

However, there is some literature on securities impact on bank performance, but this literature is focused on the new product innovations, such as mortgage backed securities (MBS), impact on economic downturns, such as the financial crisis of 2007-2008, and through it, impact on bank performance. Crotty (2009) discusses that the new financial architecture (NFA), a part of which MBS are, is the reason for the 2007-2008 financial crisis. According to Crotty, MBS were highly complex and nontransparent instruments and there was no way of pricing them correctly, which lead to the boom in the market. Gennaioli, Shleifer & Vishny (2012) discuss whether there are risks that might be unintentionally neglected while using securities. They suggest a new approach where the neglected low probability risk plays the key role in order to diminish financial fragility in the market. Bonner, Streitz & Wedow (2016) tests how ABS issuance correlates with bank loan supply and find that it had a positive correlation before the crisis and a negative one afterwards. Antoniades (2015), however, finds no evidence that holding mortgage- backed securities had impact on higher rate of smaller bank failures during the crisis, they only impacted the larger banks.

All in all, the literature reviewed only deals with impact on bank performance during the crisis, even almost a decade past the crisis. In a sense, securities are an external product for the bank, as they are (usually) not the issuer. No literature could be found regarding securities impact FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 24 on banks during normal periods in the economy. Thus, this further confirms that there is a gap in the literature of such research.

Debit cards. Although debit cards are one of the most popular means of payment with

46,7% of global non-cash payments and 70,5% of total card transaction in 2015 (World

Payments Report, 2017), there is not much research based solely on debit cards. On the other hand, the wide usage of debit cards, especially, when compared – 0,97$ trillion in value and 25bn in numbers in 2007 vs. 2,56$ trillion in value and 69,5bn in numbers in 2016 (The Federal

Reserve payment study, 2016), which is almost triple during 9 years – clearly shows the movement of banking industry to the digital age. It also clearly raises a question of the impact on performance due to this rapid growth and movement from cash to digital money.

However, debit card research is almost solely focused on the demographics of users; there is some research on the circumstances of debit card use; there is no significant research on impact of debit card innovation on bank performance.

The variety of research on the demographics of debit card users is wide. Stavins (2001) finds that debit card use is negatively correlated with age and net worth of an individual but positively correlated to education, marital status, business ownership or being a white collar worker. Similar researches using different data sources and different time periods are conducted by Klee (2006), Mantel & McHugh (2001), Borzekowski & Kiser (2008) and many more. Not surprisingly, these researches provide similar results to the one revealed by Stavins. Hayeshi &

Klee (2003) also find that debit cards are mostly used for smaller, day-to-day purchases like grocery shopping, gas or eating out or at self-service places. Even though the research on debit cards is not related to bank performance, if it is assumed that debit card usage is related to banking digitalization, the demographics of users reveal that consumers with higher social status FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 25 (i.e. education, marital status, business ownership, income etc.) are more prone to use digitalized products (e.g. in comparison to more traditional forms of money, such as hard cash). As bank performance relies on demographics of its users (Rao, 2008), it is important for traditional financial intermediaries to take into account the preferences of consumers in order to keep lead in the digitalizing financial world.

As mentioned, no significant research can be found on debit card impact on bank performance. Moreover, no significant research can be found on any consequences of debit card innovation on the overall economy, although they have been around since the 60s and were one of the first steps to a more digitalized banking.

Pre-paid cards. Prepaid cards are divided into two types – closed and open system.

Closed system prepaid cards are those issued by a specific retailer (e.g. store gift cards) and can only be used at that retailer. Open system prepaid cards are those issued by a payment network

(e.g. Visa) so they resemble debit/credit cards and can be used more flexibly - to pay for goods and services or even withdraw money from an ATM.

Cheney & Rhine (2006) divide prepaid open system cards into two groups – payroll and general spending cards. Payroll cards are used mainly for people who do not have a traditional bank account. They provide a safer, more convenient and cheaper way for them to access their salaries than cashing out checks, as well as save money for the employer on payroll check costs.

Payroll cards were fundamentally issued by non-bank financial organizations, but according to

Cheney & Rhine, “open-system payroll cards spurred several large card-issuing banks to develop programs of their own”. Banks have seen an opportunity, as well as the risk of competition, and used it to their own advantage by implementing the new product themselves. They are an example of traditional bank employing competition to their own advantage by implementing the FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 26 innovation themselves Pre-paid payroll cards are another step to a more digitalized bank, a product that helped in changing the landscape of traditional banking.

General spending cards are issued by non-bank service providers, they can be applied for over the internet, the phone or at the service provider’s location, if such exists. These cards can be topped up in several ways, such as electronic transfers, account-to-account transfers or even paper checks, if the service provider has a physical branch (Cheney & Rhine, 2006). General spending cards, similarly as payroll cards, were also adopted by banks, but there still exists non- bank service provides who issue them. Although, non-bank service providers tend to grow into banks over time, usually, digital banks (e.g. Revolut, which has applied for European banking license in 2017), which can pose as higher competition for traditional banks as they start providing more services than only pre-paid cards.

Hunt et al. (2012) use transaction data to analyze the usage of pre-paid cards. They conclude that mostly prep-paid cards are short-lived products which are used for less than six months. Although, the cards that are used longer, are used more intensively. They also find, that pre-paid cards are used both for purchases of goods and services and cash withdrawals, though cash withdrawals mostly occur in the payroll cards.

Some critique on prepaid cards are that they are too loosely or not at all regulated.

Furletti & Smith (2005) discuss that pre-paid cards lack regulation, thus, they are more prone to frauds, errors and disputes, though card issuers have extended some safeguards (e.g. zero liability). Several authors (Sienkiewicz, 2007; Choo, 2008; Simser, 2012) express concerns that pre-paid cards could be used as a money laundering tool, as they are loosely regulated ant the money is hardly tracked. Issues with regulations of financial innovation are not discussed in this paper, however, it could be a potential topic for further research. FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 27 There is a lack of research done on the impact of pre-paid cards for banks. There is some evidence in the research, such as mentioned by Cheney & Rhine (2006), that banks have adopted the innovation of pre-paid cards, but as there are independent non-bank institutions, which issue pre-paid cards, the research on their impact to bank profitability and performance is valid.

However, the fact that both, open and closed system pre-paid cards were mostly overtaken by traditional banks leads to believe that this innovation might at first show a negative effect on traditional banking performance, however, in the long run, would serve as a complementary service to already existing traditional bank services and demonstrate a positive effect on traditional bank performance.

New processes

Schumpeter (1983) has described process innovation as the establishment of a new process in the company that has not yet been used in that particular company. In this thesis, a new process is perceived as a financial innovation if it is new to the whole financial sector. The main topics for research in the new process area in finance are automated clearinghouse, risk management, asset securitization and internet banking.

Literature on asset securitization is closely related to the literature on securities, as the process itself emerges from securities. To put in simple words, asset securitization is a process of taking any illiquid or less liquid asset and, through financial engineering, turning it into a security. These securities are called asset-backed securities (ABS), which have been briefly discussed in the new product innovation part of this thesis. As neither asset securitization, nor risk management are closely related to the topic of the thesis, these subjects are considered as out of scope and are not be further discussed. The following section is focused on automated clearing houses and internet banking. FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 28 Automated clearinghouse. Automated clearinghouse (ACH) is an electronic system that deals with payroll, tax payments, consumer bills and other automatic payment services by electronically transferring the funds. ACH were invented in the 70s, but they started gaining magnitude in the 80s. In 2015, automated clearinghouse payments in the US made up 24bn in transaction volume with $41,6 trillion in value (NACHA, 2016).

A big part of literature about automated clearinghouses is focused on understanding the supply and demand conditions of the process (Bauer & Hancock, 1995; Bauer & Ferrier, 1996;

Stavins & Bauer, 1999). More recent articles, such as of Gowrisankaran & Stavins (2004) and

Ackerberg & Gowrisankaran (2006) study the network externalities for automated clearinghouses. Gowrisankaran & Stavins find that technological advancement, peer-group effects, economies of scale, and market power can all be causes of significant network externalities for automated clearinghouse. Ackerberg & Growisankaran add to the discussion by studying the hurdles for automated clearinghouse use and identify large fixed costs of bank adoption as one of the biggest of them. Peterson (2007) looks for ways of how automated clearinghouse processes could be simplified to keep its competitiveness. Peterson suggest “Four elimination” strategy that would simplify automated clearinghouses by eliminating unnecessary

SEC codes, written authorization for e-checks, opt-outs and check conversion exceptions.

Kamback & Miner (2009) prove that automated clearinghouses have not lost their position in the market with their article on international automated clearinghouse growth. They predict that electronic payment volumes should double to more than 420bn transactions by 2010. World

Payment report (2017), however, shows that that volume was reached in 2014-2015 with 433bn transactions. The authors also discuss other drivers for the interest in international automated clearing houses like the decrease of cash transactions and growth of non-cash transactions, which FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 29 also contributes to the assertion of digitalization of the financial sector. According to Esselink &

Hernandez (2017), cash transactions do not decrease in most of the world, but most of the cash transactions are for day-to-day small items, while non-cash transactions are used for bigger purchases. The value of cash and non-cash transactions in Europe are very similar though the volumes highly differ (cash volume – 129bn transactions, value – €1653bn; non-cash volume –

33bn transactions, value – €1315bn). In the context of banking performance this means that costlier purchases are made through digital means of payment, thus, bigger quantities of money come through them. As cash transactions are used for small purchases and bigger purchases are made with non-cash (digital) transactions, it means that smaller amounts of hard cash are drawn from ATMs and deposit volumes should be bigger. As mentioned, ACH’s are used for automated electronic transactions, mostly for such as loan payments, tax payments, payroll or B2B payments. ACH system can be called a pioneer for electronic money innovations, so it is highly associated with the digitalization of financial sector.

Internet banking. According to Oxford dictionary (n.d.), internet banking is “a method of banking in which transactions are conducted electronically via the Internet”. Internet banking is considered to be one of the biggest innovations for the banking sector. Mostly, research on internet banking concentrates on the characteristics that encourage internet banking implementation and/or how implementation of internet banking impacts bank performance.

The decision to implement internet banking is highly associated with several factors.

Mainly, size, age and the location of the bank. Research finds, that younger, larger and more located in urban areas banks tend to implement internet banking quicker (Sullivan, 2000; Furst,

Lang & Nolle, 2002; Nickerson & Sullivan, 2003). This confirms the idea discussed previously that consumers with higher social status are more prone to digitalization. Other factors associated FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 30 with internet banking implementation are country-level demographics and market concentration.

Banks of countries with higher income, higher education and better internet access tend to implement internet banking quicker (Hernandez-Murillo et al., 2010), which also contributes to the commemorated idea.

DeYoung, Lang & Nolle (2007) studies the impact of internet banking implementation in

1999-2001 in the United States. They purposefully choose an older time period to be able to make a clean comparison, without the impact of any other innovations. What they find is that internet banking increases bank revenues through deposit service charges, increases deposit movement from checking to money market deposit accounts and increases bank employee wages. In the context of innovation impact on traditional bank performance this would mean that internet banking increase bank performance measures which include deposits and revenues.

Ciciretti, Hasan & Zazzara (2009) compare the performance of Italian banks which have adopted internet banking to those who have not and find a strong positive correlation between adoption of internet banking and the bank’s profitability and weak correlation with risk. This implies that internet banking highly increases bank profitability and decreases risk at a lower level. Hernando

& Nieto (2007) also support the idea of profitability by stating that Spanish banks have also experienced the benefits of online banking adoption through lower costs and higher profitability.

However, even though DeYoung, Lang & Nolle ‘s (2007) research suggests that even though internet banking has a positive effect on bank performance, it is used as a complementary rather than a main mean of banking, at least for the time being. This is also acknowledged by

Hernando & Nieto (2007) for European banks.

Further research has been done in the field concentrating more on the demographics of adopter and non-adopter of online banking model. Factors that motivate the majority of the FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 31 consumers to adopt internet banking are ease of use, usefulness and trust. Trust is one of the main factors that build relationships between customers and financial institutions, whether it is a business or interpersonal relationship (Pritchard et al., 2013). Many authors agree that trust is also one of the main motivators for customers to adopt internet bank usage and continue using it moving forward (Thakur, 2014; Chemingui & Iallouna, 2013; Akhlaq & Ahmed, 2013). The user will not adopt any new product or service if he/she has low trust of the financial intermediary.

For some users, internet banking is itself an easier approach to payments, as its perceived usefulness is higher than other alternatives (Kesharwani & Bisht, 2012; Hanmer-Lloyd et al.,

2010) but the ease of use of the particular banks website is also important, especially when considering the competition between various financial intermediaries. Customers satisfaction is closely related to the convenience of banks website (Liébana-Cabanillas et al., 2013). All these factors contribute to the usage and growth of internet banking, add to the digitalization of the financial world and stimulate the development of new organizational form, such as internet–only

(digital) banks.

Though internet banking is more adopted and growing, no more recent research can be found on its impact on traditional bank performance. As the financial world moves on to a more digital landscape, the situation of internet banking might be changed from complementary to main form of banking and a literature gap on this topic can be seen. The emergence of internet banking has also promoted a new organizational form in the financial sector – digital banks, which are discussed in the following chapter.

New organizational form

Since the 80s, there has been a great rise in new forms of business organization seen. The financial sector, even though it is a highly conservative, is no exception. Security affiliates is one FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 32 of the biggest new forms of business but because this innovation was encouraged purely by regulatory changes, which are out of scope for this thesis, this organizational form is not discussed. Another major change in the organizational form of financial sector are digital banks, which, as mentioned, were greatly prompted by the growth of internet banking. For the clarity of the reader it is important to emphasize the difference between internet banking and digital bank.

As discussed before, internet banking allows customers to perform a range of financial transactions through the internet. As it allows customers to use an additional channel of banking, it is considered to be a process innovation. Digital bank is an organizational form innovation, as it is a type of bank, rather than a type of banking channel, which provides financial services solely through the internet without operating branches.

Digital banks were considered a failed innovation as most of banks that went in the digital direction later on gave it up, thus, not much research is done in the field. However, with

FinTech sector growth of 201% in 2014 (Dickerson, Skan & Masood, 2015), digital banking is gaining momentum again. In this section, the concept of digital banks and the scarce literature on the topic is reviewed.

Digital banks. Digital banks are an innovation prompted by the growth of internet usage over the world and some of the financial innovations discussed before, such as the growth of

ACHs and payment cards, which prompted the use of digital money rather than hard cash.

According to Worldbank (2017), in the 1990 0,05% of the population used the internet. In 2000, this number reached 6,77% and 28,85% in 2010 and in 2016 the part of the population using the internet was 45,91%. With such growth rates, the natural move for the banking industry seems to be a digitalized version of its services. FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 33 Digital banking research can be split up into pre-financial crisis literature and post- financial crisis literature. DeYoung (2001; 2005) has greatly contributed to the pre-financial crisis research on internet-only banks in the US. DeYoung (2001) presents the theoretical advantages of internet-only banks to both, customers and banks themselves. For the bank, the biggest advantage is cost-saving as there’s no need to operate branches. The customer should benefit from the convenience of usage and also from higher interest-rates which digital banks should be able to offer with the savings from overhead costs. Thus, digital banks should be able not only to offer better conditions for their users but also have higher growth capabilities than traditional banks. The author emphasizes that these advantages are purely theoretical and the reality might be different, although it cannot be estimated absolutely correctly as digital banks are a recent innovation. In his empirical research, DeYoung finds that digital banks perform worse than traditional banks of similar age and circumstance. Also, he finds that overhead costs are not necessarily lower in digital banks as well as interest rates are not necessarily higher but digital banks indeed grow faster than traditional banks. In his further research, DeYoung (2005) once again confirms that digital banks grow faster and corrects himself that at times, digital banks are able to offer better interest rates for customers. However, he shows that at the end, digital bank profitability is lower than the one of traditional banks of the same size and age and most of the start-ups that take on digital bank strategy, abandon it. However, the author emphasizes that the time period is too short to make any hard judgements. DeYoung also adds that even though digital banking lacks lending capabilities, it is a logical continuation of the future of banking, although, its market share might be limited as the best strategy for these kinds of banks is to deliver basic banking services, rather than the full range provided by traditional banks. FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 34 Delgado, Hernando & Nieto (2007) were the first ones to test whether digital banks are efficient in comparison with traditional banks in Europe. With some adjustments for the

European market, they use the same framework and methodology as DeYoung (2005). The findings are similar – digital banks have greater growth capabilities but lack lending capability even with age. Although because Delgado, Hernando & Nieto have conducted their research later than DeYoung, they can confirm, that when digital banks age, the profitability gap between them and traditional banks tends to decrease. In the light of the goal of this research, this would imply that digital banks that survive the start-up phase might be actual competitors of traditional banks as the profitability gap is decreasing and it becomes easier to compete. Cyree, Delcoure &

Dickens (2009) confirm that digital banks have lower profitability but they find that the profit efficiency is higher in digital banks. Their research also supports the idea that digital banks improve their profitability with age, although, do not fully reach the one of traditional banks.

However, the time span of the research is limited, thus this conclusion might not be stable.

Arnold & Ewijk (2011) take a deeper look into the critique of the stability of digital banks. Before the credit crisis, this issue was not discussed but during the crisis, it came to light.

The authors discuss several issues. Easy scalability, which is one of the advantages of digital banks, allows to capture the market share in the savings market more quickly but it raises the issue of keeping up with the lending. However, capturing the market share more quickly might be one of the competitive advantages of digital banks against traditional banks, as they can increase their customer base (by attracting new customers or taking over the ones of traditional banks) more dynamically. Arnold & Ewijk state that growth management is one of the biggest issues in digital banks. The second issue is the maintenance of depositors’ confidence, as it seems to decrease when there is no physical presence of the bank. In the context of this research, this FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 35 would imply that even though customers are attracted more quickly, it might be harder to keep them, thus it might not affect the performance of traditional banks significantly. Also, authors discuss that digital bank clients are very interest-sensitive, which brings a lot of instability for the bank in case of a downturn. This would mean that in the period of financial crisis, digital bank impact on traditional bank might decrease as they would lose one of the main competitive advantages – providing better interest rates.

The post-crisis literature treats digital banking as a part of the growing FinTech industry and rarely separates this one field in research. However, there are discussions arising about the impact of FinTech. According to PwC Global FinTech survey (2016), consumer banking and fund transfer and payment sectors are the ones that will be affected most by emergence of

FinTech. Moreover, 83% of respondents from traditional banks believe that there is a risk for their business to lose parts to FinTech companies in the future, which goes along with the idea of possible digital bank impact on traditional bank performance. Nonetheless, FinTech companies are a recent innovation, so there is not much data available and therefore most of the post-crisis literature is descriptive and emphasizes FinTechs ability to interfere with the financial industry.

Only one empirical research could be found which studies the impact of FinTech start-ups on retail bank share price by Li, Spigt & Swinkels (2017). The authors treat start-up funding as an indicator of their value and examine how it impacts stock returns of 47 banks in the USA during the period of 2010-2016. They find that funding of digital banking start-ups has a positive effect on traditional intermediaries’ stock returns which in unexpected. Dickerson, Skan & Masood

(2015) distinguishes two possible scenarios for the future – one, where FinTech disrupts the industry and banks compete for a diminishing share of the market and the other where banks collaborate with FinTech companies for a better customer experience with the knowledge from FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 36 both worlds. The second view is confirmed by Jun & Yeo (2016) in their research on the entry of

FinTech firms and competition, where they accent how FinTech firms complement, rather than compete with traditional financial intermediaries. Juengerkes (2016) agrees with the idea, though

Ferrari (2016) argues that with the new alternatives FinTech brings to the market, they could raise a significant threat of competition to traditional banks. With such varying views in descriptive literature and scarcity in empirical literature, a quantitative analysis seems necessary to shine some light on the question of financial innovation impact on traditional financial intermediaries.

Summary of literature review

The literature review part has brought some light on the landscape of financial innovation literature. It is clear that financial innovation has not received as much attention as innovation in other fields, even though, the importance of it is apparent, as both topics separately – innovation and finance – spark discussions.

In regards to previous research on innovation and financial innovation, in this thesis, financial innovation has been defined as new or significantly improved product, process, or organizational form developed for/by the financial sector, which is new to the implementing body (firm/economy) and brings better value to it or the consumer. Moreover, for the simplicity of the review, financial innovation was grouped to three main categories – product, process and organizational form.

In the product innovation section, securities, debit cards and pre-paid cards are discussed as one of the most important topics. Financial innovation is often perceived as securities innovation thus a big part of financial innovation literature is built on this topic. It focuses on risk sharing, asymmetrical information reduction and increasing liquidity. Even though security FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 37 innovation is the most extensively researched field in financial innovation, the gap of research on the impact on traditional financial intermediary’s performance of this innovation is clearly seen when reviewing the literature. Moving forward with debit and pre-paid card innovation, which from a point can be perceived as one of the starting points for financial sector digitalization as these innovations promoted the use of digital money rather than hard cash, the same research gap is seen. Most research is focused on the demographics of the users of these products or the habits of usage, which implies that consumers with higher social status are more prone to digitalization.

Debit cards are an internal bank product, as well as are part of the issued pre-paid cards.

Although, the lack of research on the impact on bank performance of these financial vehicles is surprising.

In the process innovation section, automated clearing houses (ACHs) and internet- banking are discussed. ACH innovation is a stepping stone in the digitalization of the banking world, as it simplified and prompted the use of non-cash transactions. Internet-banking innovation followed naturally with the growing usage of non-cash transactions. Most literature regarding internet banking deals with the adoptability of the technology, both form the bank and customer side, which also agrees with the idea that customers with higher social status are more prone to digitalization. There is a range of literature focusing on internet banking impact on banks performance, although, none of the found literature is recent. With technologies rapidly changing and the finance industry becoming a more and more digitalized sector, there is a gap for more recent studies on technological innovation impact on the banks performance.

The new organizational forms innovation flows naturally from both, product and process innovations, connecting the innovations from both of these parts to new ways of conducting business. Digital banks are an innovation prompted by the growth of non-cash transaction, more FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 38 automated processes in the financial sector and the banking industry moving forward to a more digital world. The literature on digital banking can be separated to pre-crisis and post-crisis. The pre-crisis literature focuses on the comparison of the profitability and performance of digital banks and traditional banks while post-crisis literature does not separate digital banking from whole FinTech innovation and is centered on more descriptive research and discussions on the impact of FinTech on traditional financial intermediaries and the whole financial sector. As pre- crisis literature mainly concludes that digital banks are not as profitable and riskier than the traditional ones and there is no significant quantitative research conducted in the post-crisis literature, although the question of impact still prevails, lack of empirical research can be identified.

After reviewing the existing literature on various financial innovation, concentrating most on its impact on traditional banking, it is clear, that there is a need for a research that shows how the digitalization of the financial sector impacts traditional financial intermediaries. Traditional banks are the cornerstone of the financial system; however, digitalization of the financial sector is inevitable. Digital banks are perceived as direct competitors of traditional banks in some research while seen as complementary services in other, thus, an analysis of the impacts they have on traditional banks may answer the question of strengths and weaknesses of both of these financial institutions and shed some light on the future of banking. Thus, naturally, the raised research question “what is the impact of digital banks on the performance of the traditional banks” is well grounded.

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 39 Research methodology

To answer the research question and reach the research goal, which is to analyze the impact of financial innovation, such as digital banks, on the performance of traditional financial intermediaries, such as traditional banks, several steps are followed:

1. Factors which are used to evaluate traditional bank performance and digital bank

measures, which might impact them, are determined;

2. Economic hypotheses are constructed in accordance to the identified measures to

clearly define the purpose and direction of the research;

3. An appropriate econometric model which would assist in examining the constructed

hypotheses is justified and econometric hypotheses are raised;

4. The scope and scale of the traditional and digital bank data set used in the research is

described;

5. Methods for data testing and modification in the selected model are clarified.

The research methodology part of this thesis follows the steps discussed above.

Performance measures

As mentioned in the literature review, the previous research on digital bank performance was focused on testing how digital banks perform in comparison with traditional banks

(DeYoung, 2005; Delgado, Hernando & Nieto, 2007; Cyree, Delcoure & Dickens, 2009). In this thesis, the main goal is to estimate whether there is an impact of digital banks on the performance of traditional banks and if so, determine whether the impact is negative or positive, rather than to compare performance. However, previous literature is used as a benchmark to define measures that are used to evaluate performance, which are relevant to both, digital and traditional banks. FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 40 Generally, performance ratios are classified to liquidity (risk), solvency (leverage), activity and profitability ratios (White, Sondhi & Fried, 1997), however as banks are not ordinary business units and the financial statements that they have to submit to the authorities differ from those of corporations, performance ratios should also differ. According to ECB

(2010), bank performance is described as “capacity to generate sustainable profitability” and “the key drivers of <…> performance remain earnings, efficiency, risk taking and leverage”. Even though various authors take different approaches to measuring performance and classifying performance ratios, the main ratios can be found in all researches. DeYoung compared 17 financial performance ratios, such as ROE, ROA, loan-to-deposit ratio and more. Although

Delgado, Hernando & Nieto followed in the footsteps of DeYoung in their research for European banks, they limited the performance ratios to 8 which measure operational performance, profitability, capitalization, leverage and business activity of the banks. Cyree, Delcoure &

Dickens test 14 financial performance ratios, although, to address some disadvantages of performance ratios, such as not controlling for possible market power or “window dressing”, the authors also include measuring the profit efficiency of both, digital and traditional banks. Other research, such as research on internet-banking, also often test the performance of banks

(Ciciretti, Hasan & Zazzara, 2009; Hernando & Nieto, 2007) and uses similar performance measures as the ones used in research on digital banking. In Table 1, the performance ratios used by different authors, both, in research on internet-banking and on digital banks, are summarized. FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 41 Table 1.

Performance measures of banking in literature.

Delgado, Hernando Ratio Ciciretti, Hasan Cyree, Delcoure & DeYoung (2005) Hernando & & Nieto classification & Zazzara (2009) Dickens (2009) Nieto (2007) (2007)

IT expenses/T Net Asset growth Equity/Total assets otal commissions&fee Asset growth ratio rate average s/Total assets assets

Marketing Net Book value of expenses/T Total lending/Deposits equity/Total otal commissions and and short-term assets average fees/Total assets debt assets

Securities Book value of brokerage physical Net lending/Total commissio assets/Total assets ns/Total assets average assets Activity Staff Non-interest Non-interest expenses/T expense/Total expenses/Average otal assets assets average assets

Non-interest Trading income/Total activity/Tot assets al assets Number of FTES/Total assets Salary&benefit expense/Total assets Total deposits/Total assets Contingent Loan-loss Total loans/Total assets&liab Loan-loss reserve/Total assets ilities/Total reserve/Total assets assets Liquidity assets Current Total loans/Total

accounts& deposits FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 42

Delgado, Hernando Ratio Ciciretti, Hasan Cyree, Delcoure & DeYoung (2005) Hernando & & Nieto classification & Zazzara (2009) Dickens (2009) Nieto (2007) (2007)

deposits/To tal assets

Total Totals loans/Total loans/Total assets assets Return-on- Return on assets Return-on-assets Return-on-assets Return-on-assets assets

Return-on- Return on equity Return-on-equity Return-on-equity Return-on-equity equity Net interest Net interest Profitability Interest&fees financial margin/Tot received on Interest margin Net interest margin margin/Earning al average loans/Total loans assets assets Interest paid on Non-interest deposits/Total margin deposits Expense on Non- premises&equip Equity/Total performing Doubtful Capital adequacy ment/Total liability loans/Total loans/Total assets ratio assets assets Expense on premises&physical Non-performing Non-performing capital/Book value loans/Total loans/Total assets of assets premises&physical Solvency capital Other non- Salary&benefit interest expense/Number of expense/Total FTES assets (Total liabilities - trading&non- interest bearing liabilities)/Total interest expense

Salary&benefit Other expense/Number of FTES

Business loans/Total loans Loan Consumer structure loans/Total loans Mortgage loans/Total loans Note: compiled by the author. FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 43 From the above summary it can be clearly seen that all authors, more or less, are interested in profitability, activity, liquidity and solvency ratios. In order to assess the impact of digital banks on traditional banks, rather than to compare the performance of the two samples, the focus is directed to how changes in digital bank impact the performance of traditional banks.

Following the classification by ECB (2010), earnings, efficiency, risk taking and leverage are the measures that define bank ‘s performance, which can be associated with ratios used in previous research. Earning measures indicate banks’ ability to generate profit while efficiency measures indicate the „ability to generate revenue from a given amount of assets and to make profit from a given source of income” (ECB, 2010), both of which can be classified as profitability measures in the sense of their purpose. Most often seen ratios of profitability in research are return on assets, return on equity, net interest margin and non-interest margin. These ratios measure banks’ ability to generate profit from both, its interest-earning and non-interest earning operations as well as measure how the earned profit compares to the asset size and equity size of the bank. As the mentioned measures are comprehensive, they are used in this research as well. Risk taking measures indicate how earnings are adjusted to the risks that are taken in order to generate them, which is the same that is shown by liquidity measures. Total loans to total assets is the most often used and most generalized measure for this purpose, thus it is selected for this research as well.

Leverage measures show the ability of the bank to meet its obligations and in some sense, the expected probability of business failure, which is also measured by solvency ratios. In the previous research, non-performing loans or doubtful loans to assets were used as the most often seen measures of solvency, however, in public financial statements of banks, these are not distinguished as separate lines; so in this research, total liabilities to equity are used as the main FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 44 solvency measure. A broader description of the measures and how they are calculated can be found in Appendix 1.

Digital bank measures

According to DeYoung & Rice (2004), the main source of bank profit is not only interest income as was believed before, but non-interest income as well, as it accounts for almost half of operating profit made by banks in the USA in the time of the research (see Figure 1).

Figure 1. Noninterest and net income as a % of total operating income in U.S. commercial banking, 1970–2003. Source: DeYoung & Rice, 2004.

With such growth of non-interest income as a profit source for banks, it is consistent to state that non-interest income is one of the main performance drivers of a bank. DeYoung & Rice also find that payment services are one of the largest sources of non-interest income. Payment services bring revenue through customer activity, thus, they are highly dependent on the number of bank users. However, data of number of users or value of transactions is not available as banks FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 45 in the US are not required to report such information. On the other hand, growing number of customer can also be reflected in the growth rate of the deposits, as higher number of users is one of the main reasons for increased deposits. If this is true, deposit growth in digital banks should have a negative impact on the number of users in traditional banks and thus, on its revenue and profit.

One of the competitive advantages of digital banks is that without having to operate branches, they are able to save a substantial amount on over-head expenses and can use some of these savings to pay higher interest rates on deposits (DeYoung, 2001), which might be one more reason for customer to switch banks, as 80% of customers see their banking relationship as transactional (North America Consumer Digital Banking Survey, 2016). If this is true and digital banks can in fact offer better interest rates, the growth in deposit value in digital banks should, in theory, have a negative impact on traditional bank non-interest income. Moreover, deposits in banks are used to finance loan issuance, thus, deposit value growth in digital banks might impact loan size and interest income in traditional banks as well. This would be due to one of two reasons: customers switching from a traditional bank to a digital bank (1) or customers transferring part of their business from a traditional bank to a digital bank (2) and by doing so, increasing deposit size. Deposits growth in digital banks can influence a number of measures in traditional banks, thus, the first economic hypothesis can be stated:

H1: Deposit growth in digital banks has a negative impact on traditional bank performance

In theory, the fact that digital banks are able to save on overhead costs allows them to offer not only higher deposit rates but also lower interest rates on loans (DeYoung, 2001). As previously mentioned, customers are willing to change banks if better services or prices are FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 46 offered, thus, similarly as for deposit growth, we assume that growth in loan size for digital banks means that customers are selecting digital banks to take out loans over traditional banks, or moving part or all of their loan business from traditional banks to digital banks. This would decrease loans in traditional banks, and, in turn, decrease traditional bank performance ratios, such as the selected risk taking measure of total loans to total assets. Also, as a significant part of assets are made up of loans, decrease in loan size also affects asset. Moreover, as banks capital requirement directly depend from assets, decrease in loans and therefore, assets, affect equity.

Moreover, bank market share is most often measured by asset size (to total assets in the market)

(Bikker & Haaf, 2002). As the size of the market is limited, it is natural that growth in digital bank asset size should lead to decrease or at least slower growth of asset size in traditional banks.

As discussed before, asset size is not only a measure of market size and a variable in performance ratios, but it directly impacts capital requirements for banks, therefore, equity. Thus, both, loan size growth and asset size growth in digital banks might have an impact on various performance measures of traditional banks as well. As already stated, loans make up a big portion of bank assets, thus, the second hypothesis is stated as follows:

H2.1.: Asset size growth in digital banks has a negative impact on traditional bank performance

H2.2. Loan size growth in digital banks has a negative impact on traditional bank performance

To summarize, the impact of digital banks on traditional banks can be measured by studying the impact on traditional bank performance measures (ROA, ROE, NIM, NNIM, loans to assets, liabilities to equity) as dependent variables. Deposit size growth, loan size growth and asset size growth in digital banks are the independent variables that should, in theory, impact the FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 47 previously mentioned performance measures of traditional banks. To perform a sophisticated analysis, further on an appropriate empirical model is selected and discussed.

Research instrument selection

In most econometric researches, relationships between variables are defined by performing ordinary least squares analysis (OLS). However, OLS analysis is useful when the goal is “to make quantitative estimates of economic relationships that previously have been completely theoretical” (Studenmund, 2017) as it does not prove that causality exists. As discussed previously, no empirical research could be found that tests how measures in digital banks impact performance in traditional banks, therefore, first, the impact has to be proved. Even though there is no empirical method to prove actual or “theoretical” causality, Granger causality method can be used to test for statistical causality (Studenmund, 2017). Thus, it is selected in this thesis as the main method of testing, accompanied by OLS analysis. OLS analysis is performed on the variables which are proved to have a causal relationship. A more comprehensive description of both methods is provided below.

Granger causality

As mentioned before, Granger causality method does not prove theoretical causality, rather it proves statistical causality. The method shows an existence of cause and effect relationship between two time-series. If past periods of one time-series variable X affects another time-series variable Y, variable X is said to be the leading variable and Granger cause variable Y

(Studenmund, 2017). From this definition, economic hypotheses, discussed previously, can be converted to econometric hypothesis (Table 2). FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 48 Table 2.

Econometric hypotheses

Economic Econometric hypothesis hypothesis

H1.1. H0: Deposit growth in digital banks Granger causes traditional bank ROA H1: Deposit growth in digital banks does not Granger cause traditional bank ROA H1.2. H0: Deposit growth in digital banks Granger causes traditional bank ROE H1: Deposit growth in digital banks does not Granger cause traditional bank ROE H1: Deposit H1.3. H0: Deposit growth in digital banks Granger causes traditional bank NIM growth in digital H1: Deposit growth in digital banks does not Granger cause traditional bank NIM banks has a H1.4. H0: Deposit growth in digital banks Granger causes traditional bank NNIM negative impact H1: Deposit growth in digital banks does not Granger cause traditional bank NNIM on traditional H1.5. H0: Deposit growth in digital banks Granger causes traditional bank loans-to-assets bank H1: Deposit growth in digital banks does not Granger cause traditional bank loans-to- performance assets H1.6. H0: Deposit growth in digital banks Granger causes traditional bank liabilities-to- equity H1: Deposit growth in digital banks does not Granger cause traditional bank liabilities-to-equity H2.1.1 H0: Asset growth in digital banks Granger causes traditional bank ROA H1: Asset growth in digital banks does not Granger cause traditional bank ROA H2.1.2 H0: Asset growth in digital banks Granger causes traditional bank ROE H : Asset growth in digital banks does not Granger cause traditional bank ROE H : Asset 1 2.1 H H : Asset growth in digital banks Granger causes traditional bank NIM growth in digital 2.1.3. 0 H : Asset growth in digital banks does not Granger cause traditional bank NIM banks has a 1 H H : Asset growth in digital banks Granger causes traditional bank NNIM negative impact 2.1.4. 0 H : Asset growth in digital banks does not Granger cause traditional bank NNIM on traditional 1 H H : Asset growth in digital banks Granger causes traditional bank loans-to-assets bank 2.1.5. 0 H : Asset growth in digital banks does not Granger cause traditional bank loans-to- performance 1 assets H2.1.6. H0: Asset growth in digital banks Granger causes traditional bank liabilities-to-equity H1: Asset growth in digital banks does not Granger cause traditional bank liabilities- to-equity H2.2.1. H0: Loan growth in digital banks Granger causes traditional bank ROA H1: Loan growth in digital banks does not Granger cause traditional bank ROA H2.2.2. H0: Loan growth in digital banks Granger causes traditional bank ROE H : Loan growth in digital banks does not Granger cause traditional bank ROE H : Loan 1 2.2. H H : Loan growth in digital banks Granger causes traditional bank NIM growth in digital 2.2.3. 0 H : Loan growth in digital banks does not Granger cause traditional bank NIM banks has a 1 H H : Loan growth in digital banks Granger causes traditional bank NNIM negative impact 2.2.4. 0 H : Loan growth in digital banks does not Granger cause traditional bank NNIM on traditional 1 H H : Loan growth in digital banks Granger causes traditional bank loans-to-assets bank 2.2.5. 0 H : Loan growth in digital banks does not Granger cause traditional bank loans-to- performance 1 assets H2.2.6. H0: Loan growth in digital banks Granger causes traditional bank liabilities-to-equity H1: Loan growth in digital banks does not Granger cause traditional bank liabilities- to-equity Note: compiled by the author FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 49 Granger causality can be done on multivariate data sets. However, the data sample used in this thesis (an extensive description is provided in the data section of the thesis) consists of 53 observations. According Hair et al. (2010), the minimal number of observations for Granger causality cannot violate the ratio of 5:1 observation to variable. As discussed in the previous part, in this research, 9 variables are used. So, in order develop a Granger causality model which would include all the 9 variables with up to 4 lags for quarterly data, the minimal number of observations should be 180 (4*5*9=180obs.), which would mean 45 years of quarterly data. That amount of data is unavailable; thus Granger causality is performed for variable pairs.

In order to eliminate the impact of a limited number of variables, Granger causality with

VAR model is used:

푛 푛 푋푡 = 훼 + ∑푖=1 휃푖푌푡−1 + ∑푖=1 휏푖푋푖−1 + 휇푡 ( 1 )

Where Xt refers to traditional bank performance measures (ROA, ROE, NIM, NNIM,

RISK or SOLVENCY); Yt refers to digital bank performance measures, that might have an impact on traditional bank performance measures (ASSET_GROWTH, DEPOSIT_GROWTH or

LOAN_GROWTH); n refer to the number of lags; μ refer to error terms; t refers to time.

The number of lags, used in the VAR model is determined by performing tests of Akaike information criterion, Schwarz-Bayesian information criterion and Hannan-Quinn information criterion.

In addition, impulse response analysis is performed on the Granger pairs. Impulse response analysis shows how variable X would react to a shock of +1 standard deviation in variable Y. This information can be used to predict the direction of the relationship between

Granger pairs. FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 50 Additional tests performed in order ensure the stability of the model are described in the data testing and modification section.

Ordinary least squares regression

Granger causality method is useful to define whether a causal relationship exists between two variables, however, it does not provide the direction and size of the impact. As the economic hypotheses are that deposit, asset and loan growth in digital banks have negative impact on traditional bank performance ratios, an additional analysis is needed to test the direction of the relationship. Due to this reason, in addition to Granger causality testing, an OLS analysis is performed on the variables that show causal relationship. A few control variables have been added to the OLS regression to adjust for the economic environment. As discussed before,

Federal reserve manages money supply in the economy partly through interest rates. Changes in

Federal funds rate affect interest rates which banks offer and as a consequence, bank performance measures. Thus, a control variable is added to adjust for changes in bank performance which are a result of changes in Federal funds rate. The crisis dummy is added to the regression in order to control for the effects of the 2007-2009 financial crisis on bank performance.

yt = δ + α1xt + … + α(1+n)xt-n + α(2+n)FEDFUNDS_RATEt + α(3+n)CRISISt + εt ( 2 )

Where y refers to performance ratio of traditional bank; x refers to digital bank variable; t refers to year; n refers to number of lags; FEDFUNDS_RATE refers to Federal Fund rates;

CRISIS refers to dummy variable for 2007-2009 financial crisis; ε refers to error term.

Explanations of variables and the expected effect (if any) can be found in Appendix 1.

Additional tests performed in order ensure the stability of the model are described in the data testing and modification section. FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 51 Both, Granger causality model and the OLS regressions are ran using GRETL open- source statistical package.

Data selection

Although there is scarce research on the topic of digital banking impact on traditional financial intermediaries, the previous significant research was concentrated on the time period before the financial crisis of 2007-2008 (DeYoung, 2005; Delgado, Hernando & Nieto, 2007;

Cyree, Delcoure & Dickens, 2009). As discussed in the literature review part, post-crisis literature is descriptive and no significant research on digital banking impact on traditional banks is done, moreover, the research that is done on the impact on traditional banks does not separate digital banking from the whole FinTech sector (Li, Spigt & Swinkels, 2017). Even though the overview of previous studies showed that there is less research made in the European market, this thesis studies commercial banks, both digital and traditional, operating in the US. Digital banks in the US are more established, while most digital banks in Europe are in the start-up phase. Due to this reason there is more extensive data available for the US market as well as the uncertainty of start-up banks surviving is avoided. The banks are selected according to their asset size, however, banks operating only domestically are selected, so the biggest American banks, such as JP Morgan Chase and Wells Fargo are omitted from the research as they operate globally.

As there could not be found any significant research on exclusively digital banking after the financial crisis of 2007-2008, a clear gap of such research is indicated, thus, the research is based on the data from the year 2004: Q4 to 2017: Q4.

Traditional banks that are measured in this research have to fulfill these conditions:

1. The bank was active and fully operating by the end of 2017: in order to perform the

research on the most recent data, only fully operating banks are selected; FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 52 2. The bank bases its business on offering mainly (not exclusively) basic banking

services: in order for the financial intermediary to classify as a commercial bank it

has to base its profits mainly on basic banking services rather other financial services

such as investment banking;

3. The bank operates mainly (not exclusively) through traditional distribution channels

such as branches: in order for the bank to classify as a traditional bank, it has to base

its profits mainly on traditional distribution channels rather than on digital channels;

4. The bank operates only in the United States;

5. The bank does not operate any separate business units serving as digital banks: in

order for the research to retrieve most accurate results, traditional banks cannot

operate a separate business unit serving as a digital bank.

In the table below (table 3) a list of traditional banks by asset size, used in this empirical research is provided.

Table 3. Traditional banks by asset size Bank name Asset size (in year-end 2017, in bn., $) US Bancorp 462,04 PNC Financial Services 381,45 TD Bank 288,29 BB&T 221,64 SunTrust Bank 205,96 Chase Bank 143,8 Fifth Third Bancorp 142,19 Regions Bank 124,29 BMO Harris bank 109,38 Huntington National Bank 104,05 USAA 80,52 Comerica Bank 71,61

Total USA 17240,00 Total 2335,27 Note: compiled by author, based on data from Bloomberg terminal FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 53 According to ECB (2007), digital banks are legal entities (rather than divisions of a traditional bank under a different brand name, such as Marcus by Goldman Sachs), which mainly rely on electronic distribution channels but also can operate some branches. This definition is used in this thesis as well. The digital banks selected for the research similarly to Delgado,

Hernando & Nieto (2007), have to fulfill these conditions:

1. The bank was active and fully operating by the end of 2017: in order to perform the

research on the most recent data, only fully operating banks are selected;

2. The bank bases its business on offering mainly (not exclusively) basic banking

services: in order for the financial intermediary to classify as a commercial bank it

has to base its profits mainly on basic banking services rather other financial services

such as investment banking;

3. The bank operates mainly (not exclusively) through electronic distribution channels

such as the internet: in order for the bank to classify as a digital bank, it has to base its

profits mainly on digital distribution channels rather than on traditional channels;

4. The bank operates only in the United States;

5. The bank is a separate legal entity rather than a division of a traditional bank under a

different brand name: in order for the research to retrieve most accurate results,

digital banks cannot operate as a subsidiary of a traditional bank.

In the table below (table 4) a list of digital banks, used in the research, is provided.

Table 4.

Digital banks by asset size

Bank name Asset size (Year-end 2017, in bn., $) Charles Schwab bank 198,59 Ally bank 137,47 FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 54 Bank name Asset size (Year-end 2017, in bn., $) Capital One 119,56 Discover bank 98,73 Synchrony bank 79,17 State Farm bank 16,70 BofI USA 8,91 Nationwide bank 7,18 First Internet bank of Indiana 2,76 Radius bank 1,14

Total USA 17240,00 Total 670,21 Note: Compiled by author, based on data from Bloomberg terminal

Both, traditional and digital bank financial data, required to calculate the ratios, is collected from the Bloomberg terminal. The data is collected in USD, so no currency standardization was required. However, in order to adapt the data for the selected research methodology, averages were used. According to Martisius (2014), averaged variables are one of the most widespread forms of variables as they provide the general characteristic of the statistical sample. Averaged variable is considered to be representative for the observed sample.

Methods for data testing

In order to ensure that the selected models are robust, several tests have to be applied on the collected data.

Seasonality. As the collected data is quarterly, it is tested for seasonality. First, seasonality is checked by plotting each variable against time. If the plots lead to believe that there might be seasonality, it is tested by running an OLS with seasonal dummies and testing for significance. In case seasonality is in fact discovered, data is seasonally adjusted.

According to Granger (1979), if there are strong seasonal affects and they are not adjusted for, the discovered causal relationships might be only between the seasonal components FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 55 and not between the non-seasonal components. However, Granger also discusses that if the data is seasonally adjusted, false causal relationships between non-seasonal components might be discovered. Thus, it seems that there is no correct way of data transformation for Granger causality model and it is left for the researcher to decide. However, as OLS analysis is performed on the same data in order to define the strength and direction of the relationship, the data set used in this research is seasonally adjusted, if adjustments are needed. In order to reduce the possibility of seasonal adjustments, the data is first tested for stationarity and transformed if needed and after that tested for seasonality as differencing data sometimes can reduce seasonality.

Stationarity. The data is tested for stationarity (unit roots). As the data is time-series, augmented Dickey-Fuller (ADF) test and Kwiatkowski–Phillips–Schmidt–Shin (KPSS) test are applied to test for stationarity. If the data is determined to be non-stationary it is transformed to stationary by differencing:

푦푡 = 푥푡 − 푥푡−1 ( 3 ) The new data set is again tested for stationarity. The procedure is repeated until the data is confirmed to be stationary.

Correlation. The transformed and adjusted data is tested for correlation by generating a correlation matrix. Data testing for correlation is important for the OLS analysis as high correlation between the exogenous variables leads to collinearity issues. However, according to

Gujarati (2004), it is nearly impossible to avoid collinearity but it is important to decide what level of it is acceptable for the research. Gujarati suggests that exogenous variables which have a correlation coefficient higher than 0.8 would not be included in the regression. However, collinearity also can be detected by calculating variance inflation factors (VIFs) of the variables.

According to Hair et al. (2010), higher VIF values indicate a higher degree of collinearity FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 56 between the exogenous variables. If high collinearity or multi-collinearity exists, it might affect the significance of the variables in the regression. As in most cases bank loans make up a significant part of assets, correlation between these two variables can be expected. However, as the OLS analysis does not include these two variables in a single regression, a correlation between loan growth and asset growth should not be an issue. Correlation of the exogenous variables is tested by constructing a correlation matrix and if a higher than 0.8 correlation is observed, VIFs are calculated and evaluated in order to verify whether significant collinearity exists.

Possible limitations of the research

In this section, the possible limitations of the selected research methodology and data are provided.

The traditional bank sample is consisted mostly of brick-and-clicks banks, meaning, that these banks use internet as a distribution channel as well, only not as the main or the only distribution channel. Thus, research on differences between digital banks and traditional banks could be biased. However, the majority of banks in the USA are either brick-and-clicks banks or digital banks, thus, there is no way to avoid this bias.

Due to the selected time period, some effects of the 2007-2008 crisis might impact the results. This might be avoided by taking a shorter period of time after the effects of crisis have completely passed. However, taking a shorter period of time would result in lack of observations to perform the research using the selected methodology. To control for the effects of the crisis, a crisis dummy variable is added to the OLS regression.

As banks are highly regulated institutions, leaving regulatory changes out of scope is another limitation of this thesis. The regulatory environment can not only affect innovation, but FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 57 also have an effect on bank performance ratios. Even though in this thesis changes in regulations are left out of scope, a recommendation for further research would be either analyzing the regulatory environment together with financial innovation impact or controlling for it in the research.

The selected performance measures of traditional banks could be considered a possible limitation of this research. First of all, performance measures overall are a snapshot of the past and do not display any evidence to believe that the same could be expected in the future. Second of all, as performance measures are a combination of various financial measures in the bank, it can be hard to identify, which exact measures are affected by the digital bank factors. Finally, the effect of changes digital bank factors on traditional bank financial measures can offset each other in combined ratios and thus, the impact could remain unnoticed. Thus, in further research it is recommended test non-combined financial measures.

A limitation of the research specific to the selected Granger causality model is data seasonality. As mentioned before, both (if seasonality exists), seasonally adjusted and seasonally non-adjusted data might show incorrect causal relationships. This pitfall is impossible to avoid in

Granger causality testing, however, one possible option is testing both, seasonally adjusted and non-seasonally adjusted data sets and comparing them. However, as this research performs both,

Granger causality analysis and OLS analysis, which requires data to be non-seasonal, only seasonally adjusted data is used in this research.

Another limitation specific to the selected model is that Granger causality, as mentioned before, does not necessarily show true or “theoretical” causality but rather shows statistical causality. The main pitfall of this model is that if both variables in the model are by any chance caused by a third variable, the null hypothesis of Granger causality might be failed to reject, even FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 58 though the initial variables do not cause each other (Maziarz, 2015). However, as OLS analysis is used on the variables that show causal relationships, this pitfall is at least partly avoided, as

OLS analysis should show no significance of the variable in this case.

One of the major limitations of this research is using digital bank growth factors instead of level data. Performing an analysis of both, impact of digital bank growth factors and impact of digital bank level data on traditional bank performance and comparing the results would be the desired approach to perform the most accurate research. However, due to time constrains of the master thesis, this approach is not selected in this particular study. There are several reasons why growth data was selected instead of level data. Previous research determines that digital banks have higher growth capabilities than traditional banks, however, there is a smaller number of banks which operate solely through digital means (DeYoung, 2001; 2005). This indicates that the volume of selected digital bank factors is smaller than traditional bank factors, thus using level data of digital banks might not have a significant econometric effect on traditional bank performance solely due to lack in volume and not due to absence of causality. Additionally, using level data would require for control variables to adjust for the economic conditions of the market.

First of all, it is very hard to determine which external economic variables could have an impact on the tested variables. Secondly, as discussed, one of the main pitfalls of Granger causality method is that the causality can be found due to a third variable causing the two tested variables and this cannot be controlled for in the model. If level data would be used, there would be a high possibility of causality due to a third variable which is not identified. Thus, either a different methodological approach should be selected in order to test level data or control variables should be cautiously selected and tested for causality with the initial variables or the robustness of the research has to be sacrificed. Due to time constraints testing for all possible control variables for FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 59 economic adjustments is impossible and sacrificing the robustness of the data is unacceptable, thus, using growth data is more compatible in this research.

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 60 Empirical research results and discussion

In this part of the thesis, the results of the conducted empirical research are discussed.

Firstly, the descriptive statistics are discussed. Then, normality tests are conducted and necessary adjustments and/or transformations are made in order to have a full picture of the data used in the research. After that, Granger causality tests are performed in order check for causality between the growth of digital bank measures and the selected traditional bank performance ratios. With the test result, hypotheses H1.1. through H2.2.6 are verified. Finally, ordinary least squares analysis is performed in order to verify the strength and the direction of the found causal relationship, if any. In the end of this chapter, the discussion on research findings is finalized by considering the mentioned findings in the light of financial innovation literature and previous research. Both, theoretical and practical implications of the finding are presented and recommendations for further research are provided considering the limitations and possible weaknesses of this research.

Descriptive statistics

The empirical research in this thesis is conducted using averaged data from 12 traditional banks and 10 digital banks for the time frame of 2004: Q3-2017: Q4. The summary statistics of the data are reviewed before any transformations or adjustments are made. The summary statistics of the variables are provided in the table below (Table 5).

Table 5.

Summary statistics of the un-adjusted data

Std. Ex. Mean Median Minimum Maximum Skewness Dev. kurtosis ROE 0.021316 0.022259 -0.04927 0.041921 0.013607 -2.8599 12.208 ROA 0.002652 0.002807 -0.00563 0.005059 0.001614 -2.8532 11.732 NIM 0.012306 0.011525 0.009634 0.020955 0.001947 2.0955 5.9103 NNIM -0.00375 -0.00386 -0.01148 0.001804 0.001969 -0.54413 4.4294 FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 61

Std. Ex. Mean Median Minimum Maximum Skewness Dev. kurtosis RISK 0.64156 0.63231 0.61011 0.6799 0.023514 0.50542 -1.4068 SOLVENCY 7.2171 7.1857 6.7361 7.9198 0.32277 0.29382 -1.2344 LOAN_GROWTH 0.043707 0.036475 -0.09029 0.2481 0.074978 1.0363 1.3434 ASSET_GROWTH 0.044689 0.036801 -0.03437 0.16559 0.039633 1.0432 1.2327 DEPOSIT_GROWTH 0.049113 0.039611 -0.02848 0.1731 0.042241 0.93404 0.57022 FEDFUNDS_RATE 0.013508 0.0019 0.0007 0.0525 0.018261 1.2062 -0.1464 Note: compiled by the author

From the summary statistics it can be seen that the skewness of all the variables is high ranging from -2.8 to 1.2 and excess kurtosis ranges from -0.14 to 12.2. Both, skewness and excess kurtosis indicate the normality of data distribution. Skewness and kurtosis values that are between -1 and 1 indicate normal distribution, while very high negative or positive values indicate that the data might not be normally distributed and requires transformation. The overview of summary statistics implies that all variables, except deposit growth and Federal

Funds rate might need transformation. The implication can be confirmed from the graphical distribution of the data (Appendix 2.) Normality tests are applied in the following sections and the needed transformations and adjustments are performed.

Normality tests

Normality tests are applied on the unadjusted data. Tests for seasonality and stationarity are applied and the needed adjustments and/or transformations are made.

Stationarity. As discussed in the research methodology part, in order to minimize the possibility of seasonal adjustments, the data is first tested for stationarity. To begin with, the original data set is tested with the ADF test. First, all variables are checked for constant and trend

(Appendix 3) and tested in accordance either with or without constant and/or trend.

Augmented Dickey-Fuller test. ROE, ROA and NIM are tested with constant; the rest of the variables are tested with both, a constant and a trend. The test results of ADF tests are displayed in Appendix 4. FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 62 The null hypothesis in ADF testing is that unit roots exist (hypothesis of non-stationarity) and it can be rejected if the p-value is < 0.05. According to ADF test, all variables are stationary

(the hypothesis of unit-root is rejected), except NIM, RISK and FEDFUNDS_RATE.

Kwiatkowski–Phillips–Schmidt–Shin test. KPSS test is performed on the same variables.

From the previous test (Appendix 3), it is known that all variables need to be tested with a trend.

The results of KPSS test are provided in Appendix 5.

KPSS test retrieves the same results as ADF test. All variables are stationary, except for

NIM, RISK and FEDFUNDS_RATE.

As both tests for stationarity retrieved the same result, NIM and FEDFUNDS_RATE variables are transformed to become stationary by differencing them by 1 and RISK is differenced by 2. After transformation, the same procedures are performed on the transformed data. Constant and trend are not used, as they have disappeared after data transformation

(Appendix 6). The results of tests on transformed data are provided in Appendix 7.

From the results on transformed data the hypothesis of unit-root can be rejected; the data is stationary.

Seasonality. As discussed in the empirical methodology section of this thesis, first, seasonality is tested by plotting the transformed variables against time (Appendix 8).

As time plot graphs for all variables are spiked, seasonal dummies are created and an

OLS regression with seasonal dummies is run to verify the existence of seasonality in the data

(Appendix 9).

The OLS regressions confirms that NNIM, LOAN_GROWTH, DEPOSIT_GROWTH and ASSET_GROWTH variables indeed demonstrate seasonality effects. Thus, adjustments for seasonality are made. X-12 ARIMA method in GRETL is used to seasonally adjust the data. This FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 63 method removes the systematic calendar effects. Time series plots of the seasonally adjusted data

can be found in Appendix 10.

Correlation. Correlation analysis is important for the OLS analysis as discussed in the

research methodology part. Correlation analysis is applied on the transformed and seasonally

adjusted data. As in the OLS regression the exogenous variables are asset growth, deposit

growth, loan growth and Federal Funds rate, a correlation matrix of these variables is constructed

and presented in Figure 2 below.

Correlation matrix 1

FEDFUNDS_RATE 1.0 -0.2 -0.1 -0.5

0.5

ASSET_GROWTH -0.2 1.0 0.8 0.5

0

LOAN_GROWTH -0.1 0.8 1.0 0.1

-0.5

DEPOSIT_GROWTH -0.5 0.5 0.1 1.0

-1 FEDFUNDS_RATE ASSET_GROWTH LOAN_GROWTH DEPOSIT_GROWTH

Figure 2. Correlation matrix of exogenous variables.

From the correlation matrix it can be observed that high correlation between asset growth

and loan growth in fact exists, as expected and discussed in the research methodology part.

However, as these two variables are not used in the same regression function, the correlation is FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 64 not an issue. Only FEDFUNDS_RATE variable is used together with other variables and this variable does not show correlation higher than 0.8, thus there is no issue with correlation.

As normality tests are completed and the necessary adjustments and/or transformations are made to the dataset, Granger causality model and OLS analysis can be performed.

Granger causality

First, to perform Granger causality test with VAR, an appropriate number of lags has to be selected. Akaike information criterion (AIC), Schwarz-Bayesian information criterion (BIC) and Hannan-Quinn information criterion (HQC) are used to identify the appropriate number of lags. As the data used is quarterly, max 4 lags are selected to perform the mentioned tests

(Appendix 11).

From the results it is seen that the recommended number of lags differ between variable pairs and AIC, BIC and HQC tests. Democracy of criterion is followed and models with all recommended lags are built in order to select the model which has less trouble in residuals with autocorrelation and normality (Appendix 12). After comparing the autocorrelation and normality of residuals test results, the selected VAR models for the Granger pairs are provided in Appendix

13. If the differences of autocorrelation and normality test results are not significantly better for a model with less lags, a model with more lags is selected instead.

After the appropriate lag selection, Granger causality tests for variable pairs are run.

Summary of test results are provided in table 6; full test results are provided in Appendix 14.

Table 6.

Summary of Granger test results

Independent F-statistic adj. r- Dependent variable p-value variable value squared Return on assets Asset growth 0.56055 0.4577 0.338309 FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 65 Independent F-statistic adj. r- Dependent variable p-value variable value squared Deposit growth 7.3138 0.0002 *** 0.572148 Loan growth 0.41246 0.5238 0.358111 Asset growth 0.5667 0.4552 0.323531 Return on equity Deposit growth 7.0999 0.0002 *** 0.552352 Loan growth 0.24662 0.91 0.245463 Asset growth 1.3512 0.2509 0.103693 Net interest margin Deposit growth 1.334 0.2739 0.094886 Loan growth 0.47178 0.627 0.060156 Asset growth 0.14007 0.7099 0.242436 Net non-interest Deposit growth 2.4223 0.1002 0.272183 margin Loan growth 0.13258 0.7174 0.242318 Asset growth 0.61712 0.608 0.591985 Loans to assets Deposit growth 0.45417 0.7158 0.587162 Loan growth 2.6892 0.0592 ** 0.64475 Asset growth 0.63296 0.4302 0.815455 Liabilities to equity Deposit growth 4.8094 0.0128** 0.835347 Loan growth 3.1823 0.0235** 0.827627 Note: compiled by the author; * - 90% confidence level, ** - 95% confidence level, *** -

99% confidence level

Granger causality test reveals that deposit growth in digital banks Granger causes ROA and ROE in traditional banks with 99% confidence level; liabilities to equity in traditional banks with 95% confidence level. Loan growth in digital banks Granger causes loans-to-assets and liabilities-to-equity in traditional banks with 95% confidence level. None of the other variables show causal relationships. As significant Granger causality is confirmed for the previously mentioned variables, hypotheses H1.1., H1.2., H1.6., H3.5. and H2.2.6. (table 2) are failed to reject; other hypotheses (H1.3. through H1.5.; H2.1. through H2.6.; H2.2.1. through H2.2.4.) are rejected.

Further analysis demonstrates the direction of the relationship between digital bank variables and traditional bank performance.

Impulse response analysis is performed next on the models that indicate the existence of a causal relationship (Appendix 15). Shock in deposit growth affects ROA after 1 quarter to some extent and then the most drastic impact is seen after 2.5 quarters. ROA returns to normal at FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 66 quarter 10. Between the initial shock and the return to normal, ROA seems to fluctuate, but has an increasing trend. ROE experiences a very similar effect of a shock in deposit growth, however, the extent to which it is affected is bigger (-0.0008 in ROA vs. -0.007 in ROE). Loan- to-asset ratio experiences a positive effect after 1 quarter of a shock in loan growth in digital banks and then a sharp negative effect after 2 quarters. It does not seem to return to normal and fluctuates between positive and negative values. After a shock in deposit growth, liabilities-to- equities in traditional banks incur a sharp positive effect after 2 quarters and then gradually returns to normal at quarter 7. However, after a shock in loan growth in digital banks, the same variable incurs a small positive effect after 1 quarter, then a sharp negative effect after quarter 2 and again, a positive effect after 4 quarters. It returns to somewhat normal at quarter 6. However, to indicate the significant direction and size of the relationship between traditional bank performance ratios and digital bank variables, an OLS analysis is performed on the variables that show causal relationships.

Ordinary least squares regression analysis

In this part, 5 OLS regressions are run. For OLS model to be robust, heteroscedasticity should not exist in the model. Thus, the formed OLS regressions are first tested for heteroscedasticity by using White’s test (Appendix 16) and as heteroscedasticity is in fact found in all the models, the models are run using GRETL function of “Heteroscedasticity corrected” model. Summarized results of the OLS analysis can be found in table 7; full results can be found in Appendix 17. FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 67 Table 7.

Summarized results of OLS analysis

Dependent Independent adj. r- Model Coefficient p-value variable variable squared Deposit growth t-1 0.0000791 0.985 Deposit growth t-2 −0.00437310 0.0884* Deposit growth t-3 −0.00953934 0.0207** OLS_1 ROA 0.500804 Deposit growth t-4 0.00158599 0.6411 FedFunds rate 0.119117 0.0682* Crisis dummy −0.000967887 0.0444** Deposit growth t-1 0.0150767 0.6575 Deposit growth t-2 −0.0127538 0.6535 Deposit growth t-3 −0.0568608 0.0997* OLS_2 ROE 0.275277 Deposit growth t-4 0.0116093 0.684 FedFunds rate 0.951963 0.1349 Crisis dummy −0.00800081 0.0245** Loan growth t-1 0.054926 0.0698* Loan growth t-2 −0.0357887 0.0095*** OLS_3 Loans to assets Loan growth t-3 0.0188642 0.2525 0.225766 FedFunds rate 0.0415059 0.9355 Crisis dummy −0.00134319 0.7643 Deposit growth t-1 2.34557 0.001*** Liabilities-to- Deposit growth t-2 2.69851 0.0003*** OLS_4 0.647909 equity FedFunds rate −9.25084 0.5281 Crisis dummy 0.172915 0.0173** Loan growth t-1 1.86799 0.0017*** Loan growth t-2 −0.0385897 0.9386 Liabilities to Loan growth t-3 0.767854 0.2912 OLS_5 0.767764 equity Loan growth t-4 2.14455 0.0009*** FedFunds rate −14.8570 0.0568* Crisis dummy 0.289808 0.0002*** Note: compiled by the author

OLS regression results confirm Granger causality test results. Both tests show that deposit growth in digital banks has an impact on traditional bank ROA, ROE and liabilities-to- equities; loan growth in digital banks has an impact on traditional bank loan-to-assets and FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 68 liabilities-to-equity. In addition, it is worth mentioning that the adjusted r-squared ratios for all the OLS models are considerably high, which indicates that the models fit the data.

Deposit growth in digital banks has a negative impact on both, traditional bank ROA and

ROE, which is also seen from impulse response analysis. If deposit growth in digital banks increase by 1%, the ROA of a traditional bank is expected to fall by 0.004% in 2nd quarter and

0.01 in 3rd quarter and the ROE of a traditional bank is expected to fall by 0.06% in 3rd quarter.

On the other hand, deposit growth in digital banks has a positive effect on traditional bank liabilities-to-equity, a solvency measure, as well as predicted by the impulse response analysis. If deposit growth in digital banks increases by 1%, liabilities-to-equity ratio in traditional banks is expected to increase by 2.35% in 1st quarter and 2.7% in 2nd quarter. Similarly, loan growth in digital banks has a positive impact on traditional bank liabilities-to-equity. The same effect was seen in impulse response analysis, however, a negative effect was noticed in Q2, which even though is seen in OLS analysis, but is not significant. The OLS analysis shows that if loan growth in digital banks increases by 1%, liabilities-to-equity in traditional banks are expected to rise by 1.87% in 1st quarter and 2.14% in 2nd quarter. However, loan growth has a negative impact on loans-to-assets in traditional banks at lag 2. Similarly, in impulse response analysis, the negative effect of a shock in loan growth is observed at Q2, while positive effects are seen at other quarters. In OLS analysis, positive relationship of loan growth in digital banks to loan-to- asset ratio in traditional banks is also seen at lags 1, 3 & 4, however, only the effect in 1st quarter is significant. The significant relationships show that a 1% increase in digital bank loan growth should increase loan-to-asset ratio of traditional banks by 0.05% in 1st quarter and decrease it by

0.03% in the 2nd quarter. The implications of these results are discussed in the next section. FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 69 Discussion

The main purpose of this part of the thesis is to summarize the results of the research, review them in the light of previous literature and discuss practical implications of the results as well as consider the implications of the limitations discussed in the research methodology part and provide suggestions for further research in the field.

In the table below (table 8) a summary of research results in the light of previously stated economic hypothesis are presented.

Table 8.

Summary of research results

Economic hypothesis Econometric model Result

Deposit growth in digital banks Granger causes traditional bank ROA, ROE and H1: Deposit growth in Granger causality SOLVENCY measures; other traditional digital banks has a bank performance ratios do not show negative impact on Granger cause relationship. traditional bank Deposit growth in digital banks has a performance negative relationship with ROA and ROE OLS analysis in traditional banks; positive relationship with SOLVENCY in traditional banks. H2.1.: Asset growth in Asset growth in digital banks has no digital banks has a Granger causality Granger cause relationship with traditional negative impact on bank performance measures traditional bank N/a OLS analysis performance Loan growth in digital banks Granger Granger causality causes traditional bank RISK and H .: Loan growth in 2.2. SOLVENCY measures digital banks has a Loan growth in digital banks has a positive negative impact on relationship with traditional bank RISK at traditional bank OLS analysis lag 1 and a negative relationship at lag 2; a performance positive relationship with traditional bank SOLVENCY. Note: compiled by the author FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 70 Granger causality test results have shown that changes in deposits in digital banks indeed have an impact on traditional bank performance, thus, H1 is failed to reject. Deposits showed to have a Granger cause relationship with traditional bank ROA, ROE and the selected measure for traditional bank solvency, liabilities to equity. As expected, OLS analysis showed that deposit growth in traditional banks negatively impacts traditional bank ROA and ROE. Because growth levels rather than level data are taken, these results indicate that if digital banks acquire deposits faster, previously mentioned performance ratios in traditional banks decrease. As stated in research methodology part, growth in digital bank deposits is considered to be a proxy for digital bank customer growth. Thus, the pace of digital bank deposit growth is assumed to be a proxy for how fast they attract new customers. The assumption discussed in research methodology part was that customers chose to deposit their funds in digital, rather than traditional banks, due to their ability to offer better conditions and this leads to customer loss in traditional banks. As deposit growth in fact was confirmed to have an impact on traditional bank performance ratios, one of the reasons for this could be the conditions they offer for the customer. Another reason for these results could be the simplicity of managing their finances solely through digital means that digital banks offer their customers. However, in order to verify these assumption, additional analysis should be performed, including a comparison of depositing conditions offered by digital and traditional banks and/or testing customer preferences of ways to manage personal finance.

DeYoung & Rice (2004) suggest that that non-interest income is a significant part of bank profitability. Net income, which is the numerator in both, ROA and ROE, is constituted of interest income and non-interest income. As the results show that deposit growth impacts ROA and ROE, in the light of the research of DeYoung & Rice, it can be assumed that this impact is caused through impact on non-interest income. However, deposits in digital banks do not show FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 71 any causal relationship with net non-interest margin of traditional banks. On the other hand, as was predicted in the limitations of this thesis, the lack of the relationship can be explained through an overview of net non-interest margin formula. The numerator of this ratio is net non- interest income, which should be decreased by growth in digital bank deposits, however, the denominator is traditional bank deposits, which also should be decreased for the same reason.

Thus, the decrease in both, numerator and denominator could offset each other and causes the lack of impact on this particular variable. A more detailed analysis on the relationships between digital bank deposits and traditional bank measures might improve and expand the results of this finding. As discussed in the limitations parts of this thesis, to verify the assumption that deposit growth causes impact on ROA and ROE through non-interest income and that there is no causal relationship with net non-interest margin due to counterbalance effect, impact on non-combined measures rather than ratios should be analyzed.

Deposit growth in digital banks also shows to have Granger cause relationship with traditional bank solvency, precisely the selected liabilities to equity ratio. However, the expected relationship was negative and the OLS analysis reveals that it is in fact positive. As discussed before, in theory, growth in digital bank deposits should decrease traditional bank deposits as customers select digital banks over traditional banks. One assumption for the positive relationship might be that deposit growth in digital banks might be not due to customers moving funds from traditional banks, but due to overall deposit growth in the market. However, if this would be the case, then ROA and ROE in traditional banks should also show positive relationship to deposit growth in digital banks. As the research shows, it is quite the opposite. As discussed in the limitations part of this thesis, bank performance ratios can also be affected by the regulatory environment, which could explain the positive relationship. Bank regulators set a FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 72 required minimum rate of leverage for banks, which means that at all times banks have to maintain an eligible ratio of leverage consistent to its assets and liabilities. If banks liabilities decrease, the required amount of equity to retain the same required leverage ratio decrease as well, thus, decrease in liabilities might also indirectly decrease equity. If this is true, then because of both variables decreasing, the ratio could increase. A deeper analysis of deposit impact on bank solvency would be required to see a broader view. There are several suggestions for further research. Firstly, comparing deposit growth in digital banks, deposits in traditional banks and the overall deposits in the market would strengthen the research. Secondly, analyzing how deposits in digital banks and deposits in traditional banks compare to solvency ratios that are set up to a certain requirement by the regulators would add credibility to the research and provide a more insightful overlook of digital bank impact on solvency measures of traditional banks.

Asset growth in digital banks has not shown Granger cause relationship with any of the selected performance measures of the traditional banks, thus H2.1. is rejected. The assumed explanation of the lack of causality between asset growth in digital banks and performance measures in traditional banks might be that assets are a complex indicator in both, digital and traditional banks. As assets are constituted of physical assets, monetary assets, loans, reserves, securities and other indicators, the growth of assets in digital banks might indeed have no cause on performance of traditional banks. To perform a more thorough analysis, digital bank assets should be broken down to more specific sections (such as trading assets, fixed assets, etc.) and then the effect of these sub-sections on corresponding performance measures of traditional banks should be measured.

Digital bank loan growth also shown to have Granger cause relationship with both, the selected risk measure of traditional banks, loans to assets and the selected solvency measure, FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 73 liabilities to equity, thus, H2.2. is failed to reject. OLS analysis has shown that the significant relationship of loan growth to loans to assets ratio in traditional banks is positive in the 1st quarter and negative in the 2nd quarter. The same effect is observed in the impulse response analysis. The impulse response analysis has shown that after a shock in digital bank loan growth, sharp fluctuations in traditional bank loans to assets ratio can be identified. One of the reasons for such results could be that, as predicted in the research limitations part, there is a third factor that influences changes in loan to assets ratio in traditional banks, that could not be identified. In further research, additional control variables might be added to the OLS analysis in order to test whether this assumption is correct. The OLS regression has also shown that loan growth in digital banks has a significant positive relationship to liabilities to equity ratio in traditional banks. As a negative relationship was the expected, this might be another point where the expected limitations of regulatory environment have influence. After the announcement of the

Third Basel Accord (Basel III), US Federal reserve has announced the minimum required leverage ratio for financial institutions, which could have influenced the positive effect of the solvency ratio. As mentioned before, controlling for changes in regulatory environment would be a meaningful improvement in further research.

As loan growth in digital banks, similarly as deposit growth, is considered to be a proxy for digital bank customer growth, it shows how fast digital banks attract new customers through increase in loans. In this sense, loan growth should decrease traditional bank net income through interest income. However, no causal relationship has been observed to ROA, ROE or NIM, as would be expected. This could be due to the limitation of testing combined ratios rather than sole financial factors. As discussed before, increased growth in digital bank loans should not only decrease traditional bank loans, but assets as well (through loans). Because of that, even though FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 74 net-income decreases through decrease in interest income, decrease in assets might offset the impact and thus, no causal relationship is observed. The same rationale can be applied to NIM – decrease in interest income is not big enough to decrease the overall ratio when loans decrease.

However, these are only assumptions. A more thorough research would provide a sound econometric answer. As in the case of deposit growth and asset growth, analyzing the level data of variables or impact on non-combined financial measures rather than ratios would be a suggestion for further research.

In order to discuss this research in the light of previous literature, it is important to note, that no similar research could be found on digital bank impact on traditional banks, thus, the review is limited. On the other hand, due to lack of previous literature in this direction, this research paper, even despite its limitations, can be considered as a foundation for further research in this direction.

In a sense, this work is an extension of research conducted before that compares digital bank performance to the performance of traditional banks, such as DeYoung (2005), Delgado,

Hernando & Nieto (2007), Cyree, Delcoure & Dickens (2009). As discussed, DeYoung (2005) finds that digital banks have better growth capabilities than traditional banks. This thesis neither confirms, nor rejects the mentioned statement, but rather gives more insight on the impact that digital bank growth capabilities have on traditional bank performance. The results of this thesis suggests that growth in digital banks poses a threat to traditional bank performance. The research performed by Li, Spigt & Swinkels (2017) is most closely related to this particular research, as somewhat they also analyze the impact of financial innovation on traditional innovation.

However, the results are completely different. In this research, the economic hypotheses stating that digital bank deposit and loan growth have a negative impact on traditional bank performance FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 75 are failed to reject, while Li, Spigt & Swinkels find a positive relationship between FinTech deal funding and traditional intermediary performance. Nevertheless, Li, Spigt & Swinkels do not distinguish digital banking from other forms of FinTech, as is done in this research. The authors proxy FinTech growth as deal funding and traditional intermediary performance as stock returns rather than financial performance measures. Thus, this paper brings a more detailed examination of the relationship between financial innovation and traditional financial intermediaries to the current literature on financial innovation.

The results of the performed research also bring more light to the question of what challenges traditional banks are facing in the age of such rapid growth of innovation. It can be confirmed that digital banks can be considered as a threat to traditional banks. The focus areas identified in this research are deposit and loan growth in digital banks, which, as discussed before, proxy for user growth in banks. These variables, more or less, impact most of the performance measures that are believed to drive bank performance. Thus, customer attraction and preservation should be one of the main goals for traditional banks in order to safeguard profit and market share moving forward.

As previously discussed both, in this part of the thesis and in the methodology part, there are several limitations to the research some of which could be improved in further research. Even though there is no way to completely separate digital banks and traditional banks, as their business models are interconnected at some level, one direction that could be followed further on is to test the relationship of other bank-like financial innovations with traditional financial intermediaries. Electronic money institutions are a growing part of FinTech innovation, however, they are not classified as banks (e.g. Revolut, PayPal). The impact of these organizational forms on traditional banks could be analyzed in further research using similar methodology. The same FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 76 methodology as in this particular thesis also could be applied in analyzing the relationship between other performance variables and other digital bank measures. A valuable extension to the research would be integrating regulatory requirements of banks, both, digital and traditional, to the research of financial measures impact. Moving forward, a completely regulatory-focused research on differences between requirements for traditional financial intermediaries and non- banking institutions providing bank-like services (such as previously mentioned electronic money institutions) could be analyzed. A different geographical zone could be selected as well as a shorter time frame excluding the 2007-2009 financial crisis could be tested, if a different methodology requiring less observations would be selected. A comparison of the impact between the US and European markets would bring more clarity to the subject in a global perspective.

Anyhow, as mentioned before, digital banking in Europe is in the start-up phase, thus such comparison would bring valid results only in the future, when more data is available.

To summarize, the performed research has failed to reject two out of three stated economic hypotheses, as deposit and loan growth indeed show impact on some performance measures in traditional banks. Thus, it can be concluded that digital banks in fact have at least partial impact on traditional bank performance. These results significantly contribute to the scarce research on financial innovation impact. In further research, several different approaches to data type, performance measures, methodology or direction of the research can be adapted to broaden the current literature even further.

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 77 Conclusions

Reviewed literature on innovation has revealed that although both, finance and innovation spark discussions in the academic field, the combined topic of financial innovation has not received as much attention from scholars as innovation in other sectors, such as manufacturing.

Although financial innovation is described differently in various sources, in this thesis it has been defined as new or significantly improved product, process, or organizational form developed for/by the financial sector, which is new to the implementing body (firm/economy) and brings better value to it or the customer. Financial innovation is grouped to three main categories – product, process and organizational form for simplicity.

The review of academic literature in these three groups has revealed that even though various financial innovations have been under the radar, most scholars concentrate on innovations that are used by financial intermediaries themselves (such as securities, ACH networks, internet banking), or dwell on descriptive rather than empirical research. Thus a gap of empirical research on financial innovation, that is not used by traditional intermediaries, on its performance has been identified.

Financial innovation, that could have the biggest impact on traditional financial intermediaries has been identified as digital banks, as they are direct competitors of traditional financial intermediaries, although, they have risen from a synergy of various other financial innovations such as ACH, pre-paid cards and internet banking.

Further analysis of the scarce empirical literature on digital bank impact on traditional intermediaries has revealed that the research focuses on performance comparison to traditional financial intermediaries rather than assessing the impact on it (DeYoung, 2005; Delgado, FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 78 Hernando & Nieto, 2007; Hernando & Nieto, 2007; Ciciretti, Hasan & Zazzara, 2009; Cyree,

Delcoure & Dickens, 2009) or fails to separate digital banking from other forms of FinTech (Li,

Spigt & Swinkels, 2017).

However, the performance ratios that were researched have been reviewed and in accordance, the ratios of traditional banks that might experience the highest impact of digital banks were selected (ROA, ROE, NIM, NNIM, RISK and SOLVENCY). Then a logical trail was followed for selecting digital bank measures that could impact those performance ratios (asset growth, deposit growth and loan growth).

Two empirical methods were selected for further research – Granger causality model, which would help identifying whether causal relationships exist and OLS analysis, which would help identifying the strength and direction of those relationships, if any.

Data required to perform the selected analysis was collected from the Bloomberg terminal and Federal Reserve database; averaged data of 12 traditional banks and 10 digital banks for the period of 2004: Q3 – 2017: Q4 was used in the analysis as well as Federal Funds rate for the mentioned periods as a control variable in the OLS analysis. The empirical research was performed using GRETL open-source statistical package.

The results of Granger causality test have revealed that deposit growth in digital banks in fact Granger causes ROA, ROE and liabilities to equity ratio in traditional banks; loan growth in digital banks Granger causes loans to assets ratio and liabilities to equity ratio in traditional banks. No other causal relationships were identified. OLS analysis has confirmed that digital bank deposit growth has a negative effect on traditional bank performance measures ROA and

ROE and a positive effect on liabilities to equity ratio; loan growth in digital banks has a positive FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 79 effect on traditional bank liabilities to equity ratio at lag 1 and a negative effect on the same ratio at lag 2 as well as a positive effect on traditional bank loans to assets ratio.

In the face of rapid changes in the financial industry due to rise of the FinTech industry, not only does this paper bring more light to the field of financial innovation impact on traditional financial intermediaries, it also reveals important implications for traditional banks. The conducted research confirms that digital banks in fact are a competitor for traditional banks. As deposit and loan growth, which are considered to be proxies for user growth, show causal relationships to traditional bank performance ratios, the research also shows that consumers should be the main focus of banks in order to safeguard high performance moving forward.

Finally, the paper builds a base for further research, that might be conducted by eliminating some of the limitations. Analyzing digital bank impact on non-combined financial measures in traditional banks rather than ratios would shine a light on the exact traditional bank factors that are impacted by digital banks. Performing an analysis of level data in addition to growth data and comparing the results might improve the accuracy of the research outcome.

Incorporation of discussion of regulatory requirements might bring some interesting implications to light as well as would testing the impact of non-bank financial innovations, such as electronic money institutions impact on traditional financial intermediaries. As usual, a different geographical region or time frame might lead to different results. A comparison of the impact of financial innovation on traditional financial intermediaries in US market and in European market would shine a light on the subject in the global perspective, however, it could only be conducted in later stages as for now, digital banking is still in the start-up phase in Europe, thus, there is no sufficient data to obtain valid results.

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 80 References

1. 2017 World Payments Report. (2017). Capgemini and BNP Paribas.

2. Accenture North America Consumer Digital Banking Survey. (2016). Retrieved from:

https://www.accenture.com/t20160609T222453__w__/us-en/_acnmedia/PDF-

22/Accenture-2016-North-America-Consumer-Digital-Banking-Survey.pdf

3. Acemoglu, D., Zilibotti, F. (1997). Was Prometheus unbound by chance? Risk,

diversification, and growth. Journal of Political Economy, 105 (4), 709-51.

4. Ackerberg, D.A., Gowrisankaran, G. (2006). Quantifying equilibrium network

externalities in the ACH banking industry. RAND Journal of Economics, 37(3), 738-761.

5. Acs, Z. J., Audretsch, D. B. (1988). Innovation in large and small firms: an empirical

analysis. American Economic Review, 78 (4), 678-690. doi: 10.1016/0165-

1765(87)90211-4

6. Ahmed, E., Akhlaq, A. (2013). The effect of motivation on trust in the acceptance of

internet banking in a low income country. International Journal of Banking Marketing,

32 (7), 115-125. doi: 10.1108/02652321311298690

7. Almarzoqi, R., Naceur, S. B., Scopelliti, A. (2015). How does bank competition affect

solvency, liquidity and credit risk? Evidence from the MENA countries. IMF Working

Papers 15/210, International Monetary Fund.

8. Anderloni, L., Bongini, P. (2009). Is Financial Innovation Still a Relevant Issue? In L.

Anderloni, D. T. Llewellyn & R. H. Schmidt (Ed.), Financial Innovation in Retail and

Corporate Banking. Cheltenham: Edward Elgar.

9. Antoniades, A. (2015). Commercial bank failures during the great recession: the real

(estate) story. ECB working paper 1779. Retrieved from: FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 81 https://www.ecb.europa.eu/pub/pdf/scpwps/ecbwp1779.en.pdf?7edad23ab5613b9693606

b94616231eb

10. Arnold, J. M. I., Ewijk, S. E. (2011). Can pure play internet banking survive the credit

crisis? Journal of Banking and Finance, 35 (4), 782-793. doi:

10.1016/j.jbankfin.2010.10.010

11. Aspara, J., Rajala, R., Tuunainen, V. K. (2012). The future of banking services. Finland:

Aalto University.

12. Bauer, P. W., Ferrier, G. D. (1996). Scale economies, cost efficiencies, and technological

change in Federal Reserve payment processing. Journal of Money, Credit and Banking,

28 (4), 1004 – 1039.

13. Bauer, P. W., Hancock, D. (1995). Scale economies and technological change in Federal

reserve ACH payment processing. Economic Review, 3 (31), 14-29.

14. Bauer, P.W., Stavins, J. (1999). The effect of pricing on demand and revenue in Federal

Reserve ACH payment processing. Journal of Financial Services Research, 16(1), 27-45.

15. Baumol, W., Faulhaber, G. (1988). Economists as innovators: practical products of

theoretical research. Journal of Economic Literature, 26 (2), 577-600.

16. Berger, A. N. (2003). Effect of technological progress: evidence from the banking

industry. Journal of Money, Credit and Banking, 35 (2), 141-176.

17. Bikker, J. A., Haaf, K. (2002). Measures of competition and concentration and

competition in the banking industry: a review of the literature. Economic & Financial FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 82 Modelling 9, Retrieved from:

https://www.dnb.nl/en/binaries/Measures%20of%20Competition_tcm47-145799.pdf

18. Bisht, S. S., Kesharwani, A. (2012). The impact of trust and perceived risk on internet

banking adoption in IndiaAn extension of technology acceptance model. International

Journal of Banking Marketing, 30 (4), 303-322. doi: 10.1108/02652321211236923

19. Blach, J. (2011). Financial innovations and their role in the modern financial system-

identification and systematization of the problem. e-Finanse: Financial Internet

Quarterly, 7 (3), 13-26. Retrieved from:

https://www.econstor.eu/bitstream/10419/66758/1/68882367X.pdf

20. Bonner, C., Streitz, D., Wedow, M. (2016). On the differential impact of securitization on

bank lending during the financial crisis. De Nederlandsche Bank working paper no. 501.

Retrieved from: https://www.dnb.nl/en/binaries/Working%20paper%20501_tcm47-

338349.pdf

21. Boot, W. A., Thakor, A. V. (1993). Security design. The Journal of Finance, 48 (4), 1349-

1378.

22. Borzekowski, R., Kiser, E. (2008). The choice at the checkout: Quantifying demand

across payment instruments. International Journal of Industrial Organization, 26(4),

889-902.

23. Cashen, L.H., Geiger, S.W. (2002). A multidimensional examination of slack and its

impact on innovation. Journal of Managerial Issues, 14(1), 68-84.

24. Cecchetti, S. C., Schoenholtz, K, L. (2015). Money, banking and financial markets (4th

ed.). Singapore: McGraw-Hill education. FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 83 25. Chandy, R.K., Tellis, G.J. (1998). Organizing for radical product innovation: the

overlooked role of willingness to cannibalize. Journal of Marketing Research 35 (4),

474–487.

26. Chemingui, H., Iallouna, H. B. (2013). Resistance, motivations, trust and intention to use

mobile financial services. International Journal of Banking Marketing, 31 (7), 574-592.

doi: 10.1108/IJBM-12-2012-0124

27. Cheney, J.S., Rhine, S.L.W. (2006). Prepaid cards: An important innovation in financial

services. Federal Reserve Bank of Philadelphia Payment Card Discussion Paper 06-07.

Retrieved from:

https://www.researchgate.net/publication/5051934_Prepaid_cards_an_important_innovati

on_in_financial_services

28. Choo, K. K. R. (2008). Money laundering risks of prepaid stored value cards. Trend and

Issues in Crime and Criminal Justice, 363. Retrieved from:

http://www.aic.gov.au/publications/current%20series/tandi/361-380/tandi363.aspx.

29. Ciciretti, R., Hasan, I., Zazzara, C. (2009). Do internet activities add value? Evidence

from the traditional banks. Journal of Financial Service Research, 35 (1), 81-98. doi:

10.1007/s10693-008-0039-2

30. Craig, J. B., Dibrell, C., Davis, P. S. (2008). Fueling innovation through information

technology in SMEs. Journal of small business management, 46(2), 203-218. doi:

10.1111/j.1540-627X.2008. 00240.x

31. Crotty, J. (2009). Structural causes of the global financial crisis: a critical assessment of

the ‘new financial architecture’. Cambridge Journal of Economics, 4 (1), 563-580. doi:

10.1093/cje/bep023 FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 84 32. Cyree, B. K., Delcoure, N., Dickens, R. (2009). An examination of the performance and

prospect for the future of internet-primary banks. Journal of Economics and Finance, 33

(2), 128-147. doi: 10.1007/s12197-008-9048-0

33. Dabrowski, M. (2017). Potential impact of financial innovation on financial services and

monetary policy. CASE Network Studies & Analyses, 488, 1-26. doi:

10.2139/ssrn.3009307

34. Damanpour, F. (1992). Organizational size and innovation. Organization Studies, 13 (3),

375–402. doi: 10.1177/017084069201300304

35. Delgado, J., Hernando, I., Nieto, M. (2007). Do European primarily internet banks show

scale and experience efficiencies? European Financial Management, 13 (4), 643-671.

doi: 10.1111/j.1468-036X.2007.00377.x

36. DeMarzo, P., Duffie, D. (1999). A liquidity-based model of security design.

Econometrica, 67 (1), 65-99.

37. DeYoung, R. (2001). The financial performance of pure play internet banks. Economic

Perspectives, 25 (1), 60-76.

38. DeYoung, R. (2005). The performance of internet based business models: Evidence from

the banking industry. Journal of Business, 78 (3), 893-947.

39. DeYoung, R., Lang, W. W., Nolle, D. L. (2007). How the internet affects output and

performance at community banks? Journal of Banking and Finance, 31 (4), 1033-1060.

doi: 10.1016/j.jbankfin.2006.10.003

40. DeYoung, R., Rice, T. (2004). How do banks make money? The fallacies of fee

income. Economic Perspectives, Federal Reserve Bank of Chicago, 28 (4), 34-51. FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 85 Retrieved from: https://www.chicagofed.org/publications/economic-

perspectives/2004/4qtr2004-part3-deyoung-rice

41. DeYoung, R., Rice, T. (2004). Noninterest Income and Financial Performance at U.S.

Commercial Banks. Financial Review, 39 (1),101–127. doi:10.1111/j.0732-

8516.2004.00069.x

42. Dickerson, J., Masood, S., Skan, J. (2015). The future of FinTech banking: digitally

disrupted or reimagined? Retrieved from: https://www.accenture.com/_acnmedia/PDF-

8/Accenture-Future-Fintech-Banking-ASG.pdf

43. Duffie, D., Rahi, R. (1994). Financial market innovation and security design: An

introduction. Journal of Economic theory, 65 (1), 1-42.

44. ECB. (2010). Beyond ROE – how to measure bank performance. Appendix to the report

on EU banking structures. Retrieved from:

https://www.ecb.europa.eu/pub/pdf/other/beyondroehowtomeasurebankperformance2010

09en.pdf

45. Esselink, H., Hernandez, L. (2017). The use of cash by households in the euro area. ECB

occasional paper series no. 201. Retrieved from:

https://www.ecb.europa.eu/pub/pdf/scpops/ecb.op201.en.pdf

46. Federal Reserve. (2016). Federal Reserve payments study 2016: developments in

consumer and business payment choices. Retrieved from:

https://www.federalreserve.gov/newsevents/pressreleases/files/2016-payments-study-

recent-developments-20170630.pdf

47. Ferrari, R. (2016). FinTech impact in retail banking – from a universal banking model to

banking verticalization. In S. Christi&J. Barberis (Eds.) The FinTech Book: the financial FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 86 technology handbook for investors, entrepreneurs and visionaries (pp 248-252). London,

England: Wiley.

48. Frame, W. S., White, L. J. (2004). Empirical studies of financial innovation: lots of talk,

little action? Journal of Economic Literature, 42 (1), 116-144.

49. Frank, R. (2015). Microeconomics and Behavior (9th ed.). New York: McGraw-Hill.

50. Fruhling, A. L., Siau, K. (2007). Assessing organizational innovation capability and its

effect on e-commerce initiatives. Journal of Computer Information Systems, 48 (1), 133–

145.

51. Fuentes, R., Hernandez-Murillo, R., Llobet, G. (2010). Strategic online banking adoption.

Journal of Banking and Finance, 34 (7), 1650-1663. doi: 10.1016/j.jbankfin.2010.03.011

52. Furletti, M., Smith, S. (2005). The laws, regulations, and industry practices that protect

consumers who use electronic payment systems: ACH e-checks and prepaid cards.

Federal Reserve Bank of Philadelphia Payment Card Discussion Paper #05-04.

Retrieved from:

https://poseidon01.ssrn.com/delivery.php?ID=43509900209212709211408309907811210

710508605403603601807600400002512201800611012602311912612210002305602009

210812409702310310205807005800407310007508710706401005209300010212709108

6000116110002022065125066088084094098027112122123112096020067004093&EX

T=pdf

53. Furst, K., Lang, W., Nolle, D. (2000). Internet banking: Developments and prospects.

Journal of Financial Services Research, 22 (1&2). FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 87 54. Gennaioli, N., Schleifer, A., Vishny, R. W. (2012). Neglected risks, financial innovation,

and financial fragility. Journal of Financial Economics, 104 (3), 452-468. doi:

10.1016/j.jfineco.2011.05.005

55. Goldsmith, R.W. (1969). Financial Structure and Development. New Haven, CT

56. Gowrisankaran, G., Stavins, J. (2004). Network externalities and technology adoption:

Lessons from electronic payments. RAND Journal of Economics, 35(2), 260-276.

57. Granger, C. W. J. (1979). Seasonality: Causation, interpretation, and implications. In A.

Zellner (ed.), Seasonal Analysis of Economic Time Series. NBER, Washington DC, 33-

56.

58. Greenwood, J., Smith, B. (1997). Financial markets in development, and the development

of financial markets. Journal of Economic Dynamics and Control, 21 (1), 145-81. doi:

10.1016/0165-1889(95)00928-0

59. Gujarati, D, N. (2004). Basic econometrics (4th ed.). McGraw Hill International.

60. Hage, J. T. (1999). Organizational innovation and organizational change. Annual Review

of Sociology 25 (1), 597. doi: 10.1146/annurev.soc.25.1.597

61. Hair, F. J., Black, C. W., Babin, J. B., Anderson, E. R. (2010). Multivariate data analysis:

a global perspective (7th ed.). New Jersey: Pearson Education.

62. Hanmer-Lloyd, S., Koenig-Lewis, N., Lifen Zhao, A., Ward, P. (2010). Adoption of

INTERNET banking services in China: is it all about trust? International Journal of

Banking Marketing, 28 (1), 7-26. doi: 10.1108/02652321011013562

63. Hayashi, F., Klee, E. (2003). Technology adoption and consumer payments: Evidence

from survey data. Review of Network Economics, 2(2), 175-190. FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 88 64. Hernando, I., Nieto, M. J. (2007). Is the internet delivery channel changing banks’

performance? The case of Spanish banks. Journal of Banking and Finance, 31 (4), 1083-

1099. doi: 10.1016/j.jbankfin.2006.10.011

65. Horne, J. C. (1985). Of financial innovations and excess. The Journal of Finance, 40 (3),

620-631.

66. Huang, D. (2015). Banks and Fintech Firms’ Relationship Status: It’s Complicated.

Retrieved from: https://www.wsj.com/articles/banks-and-fintech-firms-relationship-

status-its-complicated-1447842603

67. Hunt, R.M., van Opstal, J., Schneider, R., Wilshusen, S.M. (2012). Consumers’ use of

prepaid cards: A transaction-based analysis. Federal Reserve Bank of Philadelphia

Payment Card Discussion Paper #12-02.

68. Ireland, P. N. (1994). Money and growth: an alternative approach. American Economic

Review, 84 (1), 47-65.

69. Juengerkes, B. E. (2016). FinTech and banks – collaboration is key. In S. Christi&J.

Barberis (Eds.) The FinTech Book: the financial technology handbook for investors,

entrepreneurs and visionaries (pp 179-182). London, England: Wiley.

70. Jun, J., Yeo, E. (2016). Entry of FinTech firms and competition in retail payments market.

Asia-Pacific Journal of Financial studies, 45 (2), 159-184. doi: 10.1111/ajfs.12126

71. Kamback, E., Miner, D. (2009). Facilitating money movements globally using

international Automated Clearing House payments. Journal of Payments, Strategy and

Systems, 3 (3), 253-261.

72. Klee, E. (2006). Families use of payment instruments during a decade of change in the

U.S. payment system. Federal Reserve Board, Finance and Economics Discussion FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 89 Series, No. 2006-01. Retrieved from:

https://www.federalreserve.gov/Pubs/FEDS/2006/200601/

73. Levine, R. (1997). Financial development and economic growth: views and agenda.

Journal of Economic Literature, 35(2), 688-726.

74. Li, Y., Spigt, R., Swinkels, L. (2017). The impact of FinTech start-ups on incumbent retail

banks’ share prices. Financial Innovation, 3:26. doi: 10.1186/s40854-017-0076-7

75. Liebana-Cabanillas, S. H., Munoz-Leiva, F., Silva Bidarra, S. H. (2013). The

determinants of mobile banking acceptance: conceptual development and empirical

analysis. International Journal of Management Science & Technology Information, 8, 1-

27.

76. Mantel, B., McHugh, T. (2001). Competition and innovation in the consumer e-payments

market? Considering demand, supply, and public policy issues. Federal Reserve Bank of

Chicago, Emerging Payments Occasional Working Paper EPS-2001-4.

77. Martisius, A. S. (2014). Statistikos metodai socialiniuose ekonominiuose tyrimuose.

Vilnius: Vilniaus Universiteto leidykla.

78. Maziarz, M. (2015). A review of the Granger-causality fallacy. The Journal of

Philosophical Economics: Reflections on Economic and Social Issues, 8 (2), 86-105.

79. McKinnon, R. (1973). Money and Capital in Economic Development. Brookings:

Institution Press.

80. Merton, R. (1992). Financial innovation and economic performance. Journal of Applied

Corporate Finance, 4 (4), 12-22. doi: 10.1111/j.1745-6622.1992.tb00214.x

81. Miller, M. H. (1992). Financial innovation: achievements and prospects. Journal of

Applied Corporate Finance, 4 (4), 4-12. doi: 10.1111/j.1745-6622.1992.tb00213.x FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 90 82. Miller, M., H. (1986). Financial Innovation: The Last Twenty Years and the Next. Journal

of Financial and Quantitative Analysis, 21 (4), 459-471.

83. Mishkin, F. (1990). Financial innovation and current trends in U.S. financial markets.

U.S. – Japan Economic forum, 1, 63-77.

84. NACHA (2016). ACH volume grows by 5,6 Percent adding 1,3 billion payments in 2015.

Retrieved from: https://www.nacha.org/news/ach-volume-grows-56-percent-adding-13-

billion-payments-2015-0

85. Nelson, R.R., Winter, S.G. (1982). An evolutionary theory of economic change. UK: The

Belknap Press of Harvard University Press.

86. Nickerson, D., Sullivan, R. J. (2003). Financial innovation, strategic real options and

endogenous competition: theory and application to internet banking. Federal Reserve

Bank of Kansas City Payments System Research Working Paper PSR WP, 03-01.

Retrieved from:

https://www.researchgate.net/publication/5044855_Financial_innovation_strategic_real_

options_and_endogenous_competition_Theory_and_an_application_to_Internet_banking

87. OECD, Statistical Office of the European Communities, European Union. (1997). The

Measurement of Scientific and Technological Activities Proposed Guidelines for

Collecting and Interpreting Technological Innovation Data Oslo Manual– Oslo Manual.

Retrieved from: http://www.oecd.org/science/inno/2367580.pdf

88. Oxford dictionary. (n.d.). Internet banking definition. Retrieved from:

https://en.oxforddictionaries.com/definition/internet_banking FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 91 89. Palmberg, C. (2004). The sources of innovations - looking beyond technological

opportunities. Economics of Innovation & New Technology 13 (2), 183-197. doi:

10.1080/10438590410001628143

90. Peterson, D. (2007). How can the ACH network be simplified to compete with Check 21?

Journal of Payments Strategy & Systems, 2 (2), 197-204.

91. Pritchard, A., Roy, S., Sekhon, H., Shergill, G. (2013). Modelling trust in service

relationships: a transnational perspective. Journal of Service Marketing, 27 (1), 76-86.

doi: 10.1108/08876041311296392

92. PwC. (2016). Consumer banking survey. Retrieved from:

https://www.pwc.com/us/en/industries/financial-services/library/consumer-banking-

survey.html

93. PwC. (2016). Global FinTech Survey. Blurred lines: How FinTech is shaping Financial

Services. PwC, London. Retrieved from:

https://www.pwc.de/de/newsletter/finanzdienstleistung/assets/insurance-inside-ausgabe-

4-maerz-2016.pdf

94. Rao, S. (2008). Banking strategies and transitioning demographics. Finacle Connect, 1

(12), 10-15.

95. Ross, D., Scherer, F.M. (1990). Industrial market structure and economic performance

(3rd ed.). Boston: Houghton-Mifflin.

96. Schumpeter, J. A. (1934). Capitalism, socialism and democracy. New York: Harper and

Brothers. FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 92 97. Schumpeter, J.A. (1983). The Theory of Economic Development: An Inquiry into Profits,

Capital, Credit, Interest and the Business Cycle. New Brunswick, USA, London, UK:

Transaction Publishers.

98. Shirakawa, M. (2011). What is so special about financial innovation? Retrieved from:

https://www.boj.or.jp/en/announcements/press/koen_2011/data/ko111114a.pdf

99. Sienkiewicz, S. (2007). Prepaid cards: vulnerable to money laundering? Federal Reserve

Bank of Philadelphia Payment Card Discussion Paper #07-02. Retrieved from:

https://poseidon01.ssrn.com/delivery.php?ID=87006708811408600102808408906606810

805804501800202102310308807811311412202208608700703711411803112200210910

412708701700909405300503104408510111309807207511701006601602006500512202

3094098082090019022069065067080004106095124004099002106102085029025&EX

T=pdf

100. Simsek, A. (2013). Financial innovation and portfolio risks. American Economic

Review, 103 (3), 398-401. doi: 10.1111/j.1745-6622.1992.tb00213.x

101. Simser, J. (2012). Money laundering: emerging threats and trends. Journal of

Money Laundering Control, 16 (1), 41-54. doi: 10.1108/13685201311286841

102. Solow, R. M. (1956). A contribution to the theory of economic growth. Quarterly

Journal of Economics, Oxford Journals, 70 (1), 65–94. doi:10.2307/1884513.

103. Stavins, J. (2001). Effect of consumer characteristics on the use of payment

instruments. New England Economic Review, 3 (3), 19-31.

104. Studenmund, A. (2017). A practical guide to using econometrics (7th ed.). Pearson

Education Limited. FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 93 105. Sullivan, R. (2000). How has the adoption of internet banking affected

performance and risk in banks? Financial Industry Perspectives, 12, 1-16.

106. Thakur, R. (2014). What keeps mobile banking customers loyal? International

Journal of Banking Marketing, 32 (7), 1-6.

107. White, I, G., Sondhi, C, A., Fried, D. (1997). The analysis and use of financial

statements (2nd ed.). USA: John Wiley&Sons, Inc.

108. Worldbank. (2017). Individuals using the internet (% of population). Retrieved

from:

https://data.worldbank.org/indicator/IT.NET.USER.ZS?end=2016&start=1989&year_hig

h_desc=false

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 94 Appendices

Appendix. 1. Explanation of the variables

Expected Variable Variables Bank type Explanation Comments effect of name performance (Non-interest Data taken Net non- income-Non- Traditional from Performance NNIM interest interest bank Bloomberg measure margin expense)/Total terminal deposits Data taken Credit Traditional Total loans/Total from Performance RISK risk ratio bank assets Bloomberg measure terminal Data taken Total Leverage Traditional from Performance SOLVENCY liabilities/Total ratio bank Bloomberg measure equity terminal (Interest income- Data taken Net Traditional interest from Performance NIM interest bank expenses)/Total Bloomberg measure margin loans terminal Data taken Return on Traditional Net income/Total from Performance ROA assets bank assets Bloomberg measure terminal Data taken Net income/Total Return on Traditional from Performance ROE shareholders’ equity bank Bloomberg measure equity terminal Data taken from Bloomberg terminal; (Total Deposit Digital cumulated DEPOSIT_GROWTH depositst2/Total Negative growth bank data of deposits )-1 t1 selected digital banks used for calculations Data taken from Bloomberg Asset Digital (Assets /Assets )- terminal; ASSET_GROWTH t2 t1 Negative growth bank 1 cumulated data of selected digital banks FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 95

Expected Variable Variables Bank type Explanation Comments effect of name performance used for calculations Data taken from Bloomberg terminal; (Total Loan Digital cumulated LOAN_GROWTH loanst2/Total Negative growth bank data of loans )-1 t1 selected digital banks used for calculations Indicator taken from Fed funds Federal Reserve Control FEDFUNDS_RATE Economy Federal rate rate variable Reserve’s page Dummy variable to Control Crisis 2007: Q3 – 2010: control for CRISIS Economy dummy dummy Q4 – 1; the effects of variable 2007-2009 crisis

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 96 Appendix 2. Graphical distribution of the variables

1. Return on equity

80 Test statistic for normality: ROE N(0.021316,0.013607) Chi-square(2) = 52.404 [0.0000]

70

60

50

Density 40

30

20

10

0 -0.04 -0.02 0 0.02 0.04 0.06 ROE

2. Return on assets

700 Test statistic for normality: ROA N(0.0026522,0.0016137) Chi-square(2) = 56.998 [0.0000]

600

500

400

Density

300

200

100

0 -0.006 -0.004 -0.002 0 0.002 0.004 0.006 ROA FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 97 3. Net interest margin

600 Test statistic for normality: NIM N(0.012306,0.0019466) Chi-square(2) = 36.941 [0.0000]

500

400

Density 300

200

100

0 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02 NIM

4. Net non-interest margin

450 Test statistic for normality: NNIM N(-0.0037511,0.0019688) Chi-square(2) = 31.824 [0.0000] 400

350

300

250

Density

200

150

100

50

0 -0.012 -0.01 -0.008 -0.006 -0.004 -0.002 0 0.002 NNIM

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 98 5. Loans to assets

50 Test statistic for normality: RISK N(0.64156,0.023514) Chi-square(2) = 27.665 [0.0000] 45

40

35

30

Density 25

20

15

10

5

0 0.58 0.6 0.62 0.64 0.66 0.68 0.7 RISK

6. Liabilities to equity

2.5 Test statistic for normality: SOLVENCY N(7.2171,0.32277) Chi-square(2) = 9.385 [0.0092]

2

1.5

Density

1

0.5

0 6.5 7 7.5 8 SOLVENCY

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 99 7. Digital bank loan growth

12 Test statistic for normality: LOANG_ROWTH N(0.043707,0.074978) Chi-square(2) = 9.684 [0.0079]

10

8

Density 6

4

2

0 -0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 0.25 LOANG_ROWTH

8. Digital bank asset growth

16 Test statistic for normality: ASSET_GROWTH N(0.044689,0.039633) Chi-square(2) = 10.322 [0.0057]

14

12

10

Density 8

6

4

2

0 -0.05 0 0.05 0.1 0.15 ASSET_GROWTH

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 100 9. Digital bank deposit growth

16 Test statistic for normality: DEPOSIT_GROWTH N(0.049113,0.042241) Chi-square(2) = 10.389 [0.0055]

14

12

10

Density 8

6

4

2

0 -0.05 0 0.05 0.1 0.15 DEPOSIT_GROWTH

10. Federal funds rate

160 Test statistic for normality: FEDFUNDS_RATE N(0.013508,0.018261) Chi-square(2) = 65.289 [0.0000]

140

120

100

Density 80

60

40

20

0 -0.04 -0.02 0 0.02 0.04 0.06 FEDFUNDS_RATE

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 101 Appendix 3. Time trend and constant testing

Constant Trend p-value significance p-value significance ROE 0.0002 *** 0.4755 ROA 0.0002 *** 0.7102 NIM <0.0001 *** 0.9387 NNIM <0.0001 *** 0.0111 ** RISK <0.0001 *** <0.0001 *** SOLVENCY <0.0001 *** <0.0001 *** LOAN_GROWTH 0.0005 *** 0.0429 ** ASSET_GROWTH <0.0001 *** 0.0047 *** DEPOSIT_GROWTH <0.0001 *** 0.0294 ** FEDFUNDS_RATE <0.0001 *** 0.0002 ***

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 102 Appendix 4. ADF test results

ADF test for non-seasonally adjusted data ROE p-value 0.009121 ROA p-value 0.01117 NIM p-value 0.2982 NNIM p-value 0.0006458 RISK p-value 0.6272 SOLVENCY p-value 0.01015 LOAN_GROWTH p-value 0.04028 ASSET_GROWTH p-value 0.02932 DEPOSIT_GROWTH p-value 0.04407 FEDFUNDS_RATE p-value 0.1126

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 103 Appendix 5. KPSS test results

KPSS test for non-seasonally adjusted data

0.121 0.149 0.213 ROE Test statistic 0.143786 0.121 0.149 0.213 ROA Test statistic 0.144666 0.121 0.149 0.213 NIM Test statistic 0.221827 0.121 0.149 0.213 NNIM Test statistic = 0.201199 0.121 0.149 0.213 RISK Test statistic = 0.24362 0.121 0.149 0.213 SOLVENCY Test statistic = 0.126908 0.121 0.149 0.213 LOAN_GROWTH Test statistic 0.0786024 0.121 0.149 0.213 ASSET_GROWTH Test statistic 0.094198 0.121 0.149 0.213 DEPOSIT_GROWTH Test statistic 0.0829081 0.121 0.149 0.213 FEDFUNDS_RATE Test statistic 0.212803

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 104 Appendix 6. Time trend and constant testing for transformed data

Constant Trend p-value significance p-value significance NIM 0.9225 0.9826 RISK 0.9668 0.9656 FEDFUNDS_RATE 0.8525 0.8586

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 105 Appendix 7. KPSS and ADF test results on transformed data

ADF test on KPSS test on transformed data transformed data 10% 5% 1% 0.351 0.462 0.725 NIM p-value 7.613e-050 Test statistic = 0.145008 0.351 0.462 0.725 FEDFUNDS_RATE p-value 0.01174 Test statistic = 0.145008 0.351 0.462 0.725 RISK p-value 1.658e-018 Test statistic = 0.0655185

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 106 Appendix 8. Time series plots of the variables

1. Traditional bank return on equity, return on assets, adjusted net interest margin and

net non-interest margin

ROE ROA 0.05 0.006 0.04 0.004 0.03 0.02 0.002 0.01 0 0 -0.01 -0.002 -0.02 -0.03 -0.004 -0.04 -0.05 -0.006 2005 2008 2011 2014 2017 2005 2008 2011 2014 2017

NNIM NIM 0.002 0.006

0 0.004 0.002 -0.002 0 -0.004 -0.002 -0.006 -0.004 -0.008 -0.006

-0.01 -0.008

-0.012 -0.01 2005 2008 2011 2014 2017 2005 2008 2011 2014 2017

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 107 2. Traditional bank adjusted loans to assets and equity to liabilities

RISK 0.04

0.03

0.02

0.01

0

-0.01

-0.02

-0.03 2005 2008 2011 2014 2017

SOLVENCY 8

7.8

7.6

7.4

7.2

7

6.8

6.6 2005 2008 2011 2014 2017

3. Digital bank loan growth, asset growth, deposit growth and adjusted Federal Funds

rate

LOAN_GROWTH ASSET_GROWTH 0.25 0.18 0.16 0.2 0.14 0.15 0.12 0.1 0.1 0.08

0.05 0.06 0.04 0 0.02 0 -0.05 -0.02 -0.1 -0.04 2005 2008 2011 2014 2017 2005 2008 2011 2014 2017

DEPOSIT_GROWTH FEDFUNDS_RATE 0.2 0.006 0.004 0.15 0.002 0 -0.002 0.1 -0.004 -0.006 0.05 -0.008 -0.01 0 -0.012 -0.014 -0.05 -0.016 2005 2008 2011 2014 2017 2005 2008 2011 2014 2017

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 108 Appendix 9. OLS testing for seasonality with seasonal dummies

1. Return on assets

coefficient std. error t-ratio p-value significance const 0.00235610 0.00059504 3.96000000 0.00020000 *** time −6.18384e-06 0.00001482 −0.4174 0.67830000 dq1 0.00047096 0.00063501 0.74170000 0.46190000 dq2 0.00072297 0.00063484 1.13900000 0.26040000 dq3 0.00069394 0.00063501 1.09300000 0.27990000

2. Return on equity

coefficient std. error t-ratio p-value const 0.02001790 0.00499476 4.00800000 0.00020000 *** time −9.76988e-05 0.00012437 −0.7856 0.43600000 dq1 0.00409804 0.00533030 0.76880000 0.44580000 dq2 0.00615423 0.00532885 1.15500000 0.25390000 dq3 0.00579267 0.00533030 1.08700000 0.28260000

3. Adjusted net interest margin

coefficient std.error t-ratio p-value significance const 0.00024213 0.00056510 0.42850000 0.67030000 time −1.08083e-06 0.00001356 −0.07971 0.93680000 dq1 −0.00100463 0.00057542 −1.746 0.10740000 dq2 0.00026839 0.00057462 0.46710000 0.64260000 dq3 −0.000249626 0.00057414 −0.4348 0.66570000

4. Net non-interest margin

coefficient std. Error t-ratio p-value significance const −0.00351575 0.00088539 −3.971 0.00020000 *** time −4.35094e-05 0.00001683 −2.585 0.01280000 ** dq1 0.00130365 0.00069616 1.87300000 0.06720000 * dq2 0.00123732 0.00068903 1.79600000 0.07880000 * dq3 0.00128871 0.00065339 1.97200000 0.05430000 *

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 109 5. Adjusted loans to assets

coefficient std.error t-ratio p-value significance const −0.00104381 0.00396059 −0.2636 0.79330000 time 0.0000 0.00009637 0.02695000 0.97860000 dq1 0.00346701 0.00405178 0.85570000 0.39660000 dq2 0.00378657 0.00397347 0.95300000 0.34560000 dq3 −0.00314592 0.00396996 −0.7924 0.43220000

6. Liabilities to equity

coefficient std.error t-ratio p-value significance const 7.75128000 0.05565220 139.30000000 0.00000000 *** time −0.0183532 0.00138569 −13.24 0.00000000 *** dq1 −0.0165702 0.05939080 −0.2790 0.78140000 dq2 −0.0505722 0.05937470 −0.8517 0.39860000 dq3 −0.0904459 0.05939080 −1.523 0.13430000

7. Digital bank loan growth

coefficient std. error t-ratio p-value significance Const 0.139624 0.0215228 6.487 0.000000045 *** Time −0.000991840 0.000535897 −1.851 0.0704 * dq1 −0.125741 0.0229687 −5.474 0.00000157 *** dq2 −0.0764213 0.0229624 −3.328 0.0017 *** dq3 −0.0797030 0.0229687 −3.470 0.0011 ***

8. Digital bank asset growth

coefficient std. Error t-ratio p-value significance Const 0.0982792 0.0126675 7.758 5.11E-10 *** Time −0.000837994 0.00021949 −3.818 0.0004 *** dq1 −0.0563066 0.0110973 −5.074 0.00000626 *** dq2 −0.0465707 0.0104127 −4.473 0.0000473 *** dq3 −0.0233638 0.00883806 −2.644 0.011 ** FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 110 9. Digital bank deposit growth

coefficient std. Error t-ratio p-value significance const 0.0873243 0.0173748 5.026 0.00000738 *** time −0.000832397 0.00034738 −2.396 0.0205 ** dq1 −0.0342563 0.0160843 −2.130 0.0383 ** dq2 −0.0258879 0.0145837 −1.775 0.0822 * dq3 −0.00401374 0.0141246 −0.2842 0.7775

10. Adjusted Federal Funds rate

coefficient std.error t-ratio p-value significance const −0.00137778 0.00150036 −0.9183 0.36310000 time 0.0000 0.00003600 0.31770000 0.75220000 dq1 0.00090354 0.00152775 0.59140000 0.55710000 dq2 0.00119210 0.00152563 0.78140000 0.43850000 dq3 0.00158067 0.00152435 1.03700000 0.30510000

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 111 Appendix 10. Time series plots of seasonally adjusted digital bank asset growth, digital bank loan growth, digital bank deposit growth and net non-interest margin variables.

ASSET_GROWTH LOAN_GROWTH 0.18 0.35 0.16 0.3 0.14 0.25 0.12 0.2 0.1 0.15 0.08 0.1 0.06 0.05 0.04 0.02 0 0 -0.05 -0.02 -0.1 2005 2008 2011 2014 2017 2005 2008 2011 2014 2017

DEPOSIT_GROWTH NNIM 0.18 0.002 0.16 0 0.14 0.12 -0.002

0.1 -0.004 0.08 0.06 -0.006 0.04 -0.008 0.02 -0.01 0 -0.02 -0.012 2005 2008 2011 2014 2017 2005 2008 2011 2014 2017

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 112 Appendix 11. VAR lag selection

Granger causality mode Number of lags AIC BIC HQC

1 -14.563253* -14.254384* -14.446068* 2 -14.456967 -13.993664 -14.281191 Asset growth-->ROA 3 -14.375942 -13.758205 -14.141574 4 -14.406988 -13.634817 -14.114027 1 -10.275859* -9.966991* -10.158675* 2 -10.167048 -9.703745 -9.991272 Asset growth-->ROE 3 -10.083324 -9.465586 -9.848955 4 -10.122569 -9.350397 -9.829608 1 -14.435512* -14.123645* -14.317657* 2 -14.304295 -13.836495 -14.127513 Asset growth-->NIM 3 -14.234031 -13.610297 -13.998321 4 -14.129606 -13.349939 -13.834968 1 -14.047637* -13.738768* -13.930453* 2 -13.899091 -13.435788 -13.723315 Asset growth-->NNIM 3 -13.805565 -13.187828 -13.571197 4 -13.784902 -13.012731 -13.491942 1 -11.071058 -10.756139* -10.952552 2 -11.081277 -10.608899 -10.903518 Asset growth-->RISK 3 -11.240296* -10.610458 -11.003284* 4 -11.12242 -10.335123 -10.826155 1 -5.248738* -4.939870* -5.131554* 2 -5.119011 -4.655708 -4.943234 Asset growth-->SOLVENCY 3 -5.080028 -4.462291 -4.845659 4 -5.128324 -4.356153 -4.835363 1 -14.315833 -14.006964* -14.198648* 2 -14.315399 -13.852096 -14.139623 Deposit growth-->ROA 3 -14.269649 -13.651912 -14.03528 4 -14.413091* -13.640919 -14.12013 1 -10.03678 -9.727912* -9.919596* 2 -10.037804 -9.574501 -9.862028 Deposit growth-->ROE 3 -9.987346 -9.369609 -9.752977 4 -10.123356* -9.351185 -9.830395 1 -13.911089 -13.599222* -13.793234* 2 -13.918319* -13.450519 -13.741537 Deposit growth-->NIM 3 -13.81762 -13.193886 -13.58191 4 -13.740698 -12.961031 -13.446061 1 -13.806424 -13.497555* -13.68924 2 -13.878340* -13.415037 -13.702563* Deposit growth-->NNIM 3 -13.772293 -13.154555 -13.537924 4 -13.835811 -13.06364 -13.542851 1 -10.480191 -10.165273* -10.361685 2 -10.516346 -10.043968 -10.338587 Deposit growth-->RISK 3 -10.770990* -10.141152 -10.533978* 4 -10.719409 -9.932112 -10.423144 Deposit growth-->SOLVENCY 1 -4.696014 -4.387146* -4.57883 FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 113

Granger causality mode Number of lags AIC BIC HQC 2 -4.772249* -4.308946 -4.596472* 3 -4.645519 -4.027782 -4.411151 4 -4.579444 -3.807272 -4.286483 1 -13.251551 -12.942683* -13.134367 2 -13.256662 -12.79336 -13.080886 Loan growth-->ROA 3 -13.101544 -12.483807 -12.867176 4 -13.523618* -12.751446 -13.230657* 1 -8.963982 -8.655113* -8.846798 2 -8.967742 -8.504439 -8.791965 Loan growth-->ROE 3 -8.813134 -8.195397 -8.578765 4 -9.233606* -8.461434 -8.940645* 1 -13.043982 -12.732115* -12.926127* 2 -13.086367* -12.618567 -12.909584 Loan growth-->NIM 3 -13.018006 -12.394273 -12.782296 4 -12.904834 -12.125167 -12.610197 1 -12.763243 -12.454374* -12.646059* 2 -12.811625 -12.348323 -12.635849 Loan growth-->NNIM 3 -12.734279 -12.116542 -12.499911 4 -12.855476* -12.083305 -12.562515 1 -9.963442 -9.648523 -9.844936 2 -10.132104 -9.659726* -9.954345 Loan growth-->RISK 3 -10.201575* -9.571737 -9.964563* 4 -10.083817 -9.29652 -9.787552 1 -4.152565 -3.843697* -4.035381 2 -4.196494 -3.733191 -4.020718 Loan growth-->SOLVENCY 3 -4.111107 -3.49337 -3.876738 4 -4.604521* -3.832349 -4.311560*

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 114 Appendix 12. VAR lag model testing for autocorrelation and normality of

residuals

Expected causal Number of Autocorrelation Normality of residuals relationship lags Doornik-Hansen test 1 Durbin-Watson 2.341406 Chi-square(4) = 14.4489 [0.0060] Asset growth -> RISK Doornik-Hansen test 3 Durbin-Watson 1.968732 Chi-square(4) = 11.5211 [0.0213] Doornik-Hansen test Durbin-Watson 1 Chi-square(4) = 39.7755 2.291318 [0.0000] Deposit growth-->ROA Doornik-Hansen test Durbin-Watson 4 Chi-square(4) = 25.3073 1.832271 [0.0000] Doornik-Hansen test Durbin-Watson 1 Chi-square(4) = 47.1028 2.279408 [0.0000] Deposit growth -> ROE Doornik-Hansen test Durbin-Watson 4 Chi-square(4) = 27.9134 1.831316 [0.0000] Doornik-Hansen test Durbin-Watson 1 Chi-square(4) = 58.6396 1.897410 [0.0000] Deposit growth -> NIM Doornik-Hansen test 2 Durbin-Watson 1.927227 Chi-square(4) = 40.45 [0.0000] Doornik-Hansen test 1 Durbin-Watson 2.039642 Chi-square(4) = 40.0396 [0.0000] Deposit growth->NNIM Doornik-Hansen test 2 Durbin-Watson 1.883498 Chi-square(4) = 37.3381 [0.0000] Doornik-Hansen test 1 Durbin-Watson 2.360875 Chi-square(4) = 14.0791 [0.0070] Deposit growth -> RISK Doornik-Hansen test 3 Durbin-Watson 2.016426 Chi-square(4) = 4.16324 [0.3844] Doornik-Hansen test 1 Durbin-Watson 2.001366 Chi-square(4) = 10.0856 Deposit growth -> [0.0390] SOLVENCY Doornik-Hansen test 2 Durbin-Watson 1.967536 Chi-square(4) = 9.08373 [0.0590] FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 115

Expected causal Number of Autocorrelation Normality of residuals relationship lags Doornik-Hansen test Durbin-Watson 1 Chi-square(4) = 165.988 2.155418 [0.0000] Loan growth -> ROA Doornik-Hansen test Durbin-Watson 4 Chi-square(4) = 100.247 1.937357 [0.0000] Doornik-Hansen test Durbin-Watson 1 Chi-square(4) = 192.52 2.147874 [0.0000] Loan growth -> ROE Doornik-Hansen test 4 Durbin-Watson 1.942651 Chi-square(4) = 118.902 [0.0000] Doornik-Hansen test Durbin-Watson 1 Chi-square(4) = 94.8746 1.979492 [0.0000] Loan growth -> NIM Doornik-Hansen test 2 Durbin-Watson 1.976487 Chi-square(4) = 75.1643 [0.0000] Doornik-Hansen test 1 Durbin-Watson 1.990814 Chi-square(4) = 65.2644 [0.0000] Loan growth -> NNIM Doornik-Hansen test 4 Durbin-Watson 2.062071 Chi-square(4) = 49.2925 [0.0000] Doornik-Hansen test 2 Durbin-Watson 2.225675 Chi-square(4) = 16.5005 [0.0024] Loan growth -> RISK Doornik-Hansen test 3 Durbin-Watson 2.005202 Chi-square(4) = 15.7342 [0.0034] Doornik-Hansen test 1 Durbin-Watson 1.932224 Chi-square(4) = 28.8801 Loan growth -> [0.0000] SOLVENCY Doornik-Hansen test 4 Durbin-Watson 1.860207 Chi-square(4) = 11.3655 [0.0227]

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 116 Appendix 13. VAR models used in the research

VAR model Asset growth-->ROA VAR (1) Asset growth-->ROE VAR (1) Asset growth-->NIM VAR (1) Asset growth-->NNIM VAR (1) Asset growth-->RISK VAR (3) Asset growth-->SOLVENCY VAR (1) Deposit growth-->ROA VAR (4) Deposit growth-->ROE VAR (4) Deposit growth-->NIM VAR (4) Deposit growth-->NNIM VAR (2) Deposit growth-->RISK VAR (3) Deposit growth-->SOLVENCY VAR (2) Loan growth-->ROA VAR (4) Loan growth-->ROE VAR (4) Loan growth-->NIM VAR (2) Loan growth-->NNIM VAR (1) Loan growth-->RISK VAR (3) Loan growth-->SOLVENCY VAR (4)

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 117 Appendix 14. Full results of Granger causality tests

1. Asset growthROA

Equation 1: ROA

Coefficient Std. Error t-ratio p-value const 0.00146578 0.000766559 1.912 0.0618 * ROA_1 0.572439 0.122249 4.683 <0.0001 *** ASSET_GROWTH −0.00505546 0.00675231 −0.7487 0.4577 _1 time −4.48568e-06 1.35504e-05 −0.3310 0.7421

Mean dependent var 0.002633 S.D. dependent var 0.001624 Sum squared resid 0.000084 S.E. of regression 0.001321 R-squared 0.377232 Adjusted R-squared 0.338309 F(3, 48) 9.691768 P-value(F) 0.000041 rho −0.044977 Durbin -Watson 2.079029

F-tests of zero restrictions: All lags of ROA F(1, 48) = 21.926 [0.0000] All lags of ASSET_GROWTH F(1, 48) = 0.56055 [0.4577]

2. Loan growth  ROA

Equation 1: ROA

Coefficient Std. Error t-ratio p-value const 0.000819152 0.000599127 1.367 0.1779 ROA_1 0.622888 0.116874 5.330 <0.0001 *** LOAN_GROWTH _1 0.00234210 0.00364682 0.6422 0.5238 time 2.27596e-06 1.27968e-05 0.1779 0.8596

Mean dependent var 0.002633 S.D. dependent var 0.001624 Sum squared resid 0.000084 S.E. of regression 0.001323 R-squared 0.375327 Adjusted R-squared 0.336285 F(3, 48) 9.613417 P-value(F) 0.000044 rho −0.082764 Durbin -Watson 2.155418

F-tests of zero restrictions: All lags of ROA F(1, 48) = 28.404 [0.0000] All lags of LOAN_GROWTH F(1, 48) = 0.41246 [0.5238]

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 118 3. Deposit growth  ROA

Equation 1: ROA

Coefficient Std. Error t-ratio p-value const 0.00345012 0.00107041 3.223 0.0026 ** ROA_1 0.278151 0.147889 1.881 0.0675 * ROA_2 −0.0565259 0.157420 −0.3591 0.7215 ROA_3 −0.0684819 0.154146 −0.4443 0.6593 ROA_4 0.410935 0.127165 3.232 0.0025 *** DEPOSIT_GROW_1 −0.0147568 0.00503403 −2.931 0.0056 *** DEPOSIT_GROW_2 −0.00921735 0.00527792 −1.746 0.0886 * DEPOSIT_GROW_3 −0.0163074 0.00535062 −3.048 0.0041 *** DEPOSIT_GROW_4 0.00446043 0.00506477 0.8807 0.3839 time −2.06880e-05 1.34614e-05 −1.537 0.1324

Mean dependent var 0.002553 S.D. dependent var 0.001638 Sum squared resid 0.000045 S.E. of regression 0.001072 R-squared 0.652370 Adjusted R-squared 0.572148 F(9, 39) 8.132029 P-value(F) 1.14e-06 rho 0.123388 Durbin -Watson 1.743337 F-tests of zero restrictions: All lags of ROA F(4, 39) = 5.7943 [0.0009] All lags of DEPOSIT_GROW F(4, 39) = 7.3138 [0.0002] All vars, lag 4 F(2, 39) = 6.0149 [0.0053]

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 119 4. Asset growth  ROE

Equation 1: ROE

Coefficient Std. Error t-ratio p-value const 0.0128308 0.00653531 1.963 0.0554 * ROE_1 0.557248 0.123725 4.504 <0.0001 *** ASSET_GROWTH −0.0431875 0.0573691 -0.7528 0.4552 _1 time −5.95701e-05 0.000116686 −0.5105 0.6120

Mean dependent var 0.021140 S.D. dependent var 0.013679 Sum squared resid 0.006076 S.E. of regression 0.011251 R-squared 0.363323 Adjusted R-squared 0.323531 F(3, 48) 9.130482 P-value(F) 0.000069 rho 0.043151 Durbin -Watson 2.075561

F-tests of zero restrictions: All lags of ROE F(1, 48) = 20.285 [0.0000] All lags of ASSET_GROWTH F(1, 48) = 0.56671 [0.4552]

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 120 5. Deposit growth  ROE

Equation 1: ROE

Coefficient Std. Error t-ratio p-value const 0.0287019 0.00924673 3.104 0.0035 *** ROE_1 0.271562 0.148836 1.825 0.0757 * ROE_2 −0.0552736 0.158632 −0.3484 0.7294 ROE_3 −0.0621942 0.155292 −0.4005 0.6910 ROE_4 0.405097 0.127850 3.169 0.0030 *** DEPOSIT_GROW_1 −0.125399 0.0434254 −2.888 0.0063 *** DEPOSIT_GROW_2 −0.0771715 0.0453814 −1.701 0.0970 * DEPOSIT_GROW_3 −0.135455 0.0459519 −2.948 0.0054 *** DEPOSIT_GROW_4 0.0430183 0.0434663 0.9897 0.3284 time −0.000188914 0.000118174 −1.599 0.1180

Mean dependent var 0.020394 S.D. dependent var 0.013743 Sum squared resid 0.003298 S.E. of regression 0.009195 R-squared 0.636286 Adjusted R-squared 0.552352 F(9, 39) 7.580795 P-value(F) 2.54e-06 rho 0.123544 Durbin-Watson 1.745095

F-tests of zero restrictions: All lags of ROE F(4, 39) = 5.3749 [0.0015] All lags of DEPOSIT_GROW F(4, 39) = 7.0999 [0.0002] All vars, lag 4 F(2, 39) = 5.8657 [0.0059]

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 121 6. Loan growth  ROE

Equation 1: ROE

Coefficient Std. Error t-ratio p-value const 0.00276102 0.00849146 0.3252 0.7468 ROE_1 0.502517 0.156541 3.210 0.0027 *** ROE_2 0.0914431 0.177097 0.5163 0.6085 ROE_3 −0.0576470 0.176684 −0.3263 0.7460 ROE_4 0.185888 0.158539 1.173 0.2481 LOAN_GROWTH_1 0.0188134 0.0359228 0.5237 0.6034 LOAN_GROWTH_2 0.0118333 0.0367026 0.3224 0.7489 LOAN_GROWTH_3 0.00967335 0.0365639 0.2646 0.7927 LOAN_GROWTH_4 −0.0201279 0.0357442 −0.5631 0.5766 time 6.18320e-05 0.000146515 0.4220 0.6753

Mean dependent var 0.020394 S.D. dependent var 0.013743 Sum squared resid 0.005558 S.E. of regression 0.011938 R-squared 0.386939 Adjusted R-squared 0.245463 F(9, 39) 2.735018 P-value(F) 0.014109 rho 0.025388 Durbin-Watson 1.942651

F-tests of zero restrictions: All lags of ROE F(4, 39) = 5.8249 [0.0009] All lags of LOAN_GROWTH F(4, 39) = 0.24662 [0.9100] All vars, lag 4 F(2, 39) = 1.0225 [0.3691]

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 122 7. Asset growth  NIM

Equation 1: NIM

Coefficient Std. Error t-ratio p-value const −0.000448167 0.000616898 −0.7265 0.4711 NIM_1 −0.372815 0.132914 −2.805 0.0073 ** ASSET_GROWTH_1 0.00778820 0.00670005 1.162 0.2509 time 3.11683e-06 1.44264e-05 0.2160 0.8299

Mean dependent var −3.66e-06 S.D. dependent var 0.001482 Sum squared resid 0.000093 S.E. of regression 0.001403 R-squared 0.157472 Adjusted R-squared 0.103693 F(3, 47) 2.928156 P-value(F) 0.043263 rho -0.012556 Durbin -Watson 2.025079

F-tests of zero restrictions: All lags of NIM F(1, 47) = 7.8677 [0.0073] All lags of ASSET_GROWTH F(1, 47) = 1.3512 [0.2509]

8. Deposit growth  NIM Equation 1: NIM

Coefficient Std. Error t-ratio p-value const 2.83648e-05 0.000738490 0.03841 0.9695 NIM_1 −0.304220 0.147863 −2.057 0.0456 * NIM_2 0.0593509 0.145299 0.4085 0.6849 DEPOSIT_GROW_1 −0.00531314 0.00561682 −0.9459 0.3493 DEPOSIT_GROW_2 0.00689239 0.00536396 1.285 0.2055 time −3.73872e-06 1.54976e-05 −0.2412 0.8105

Mean dependent var −8.35e-06 S.D. dependent var 0.001497 Sum squared resid 0.000089 S.E. of regression 0.001424 R-squared 0.187245 Adjusted R-squared 0.094886 F(5, 44) 2.027366 P-value(F) 0.093303 rho 0.030678 Durbin -Watson 1.927227

F-tests of zero restrictions: All lags of NIM F(2, 44) = 2.8631 [0.0678] All lags of DEPOSIT_GROW F(2, 44) = 1.334 [0.2739] All vars, lag 2 F(2, 44) = 0.95714 [0.3918]

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 123 9. Loan growth  NIM

Equation 1: NIM

Coefficient Std. Error t-ratio p-value const −0.000239451 0.000587376 −0.4077 0.6855 NIM_1 −0.356302 0.150147 −2.373 0.0221 * NIM_2 0.0550880 0.147865 0.3726 0.7113 LOAN_GROWTH_1 0.00317513 0.00396578 0.8006 0.4276 LOAN_GROWTH_2 0.00167746 0.00398996 0.4204 0.6762 time 8.42996e-07 1.51708e-05 0.05557 0.9559

Mean dependent var −8.35e-06 S.D. dependent var 0.001497 Sum squared resid 0.000093 S.E. of regression 0.001451 R-squared 0.156058 Adjusted R-squared 0.060156 F(5, 44) 1.627261 P-value(F) 0.172765 rho 0.011467 Durbin -Watson 1.976487

F-tests of zero restrictions: All lags of NIM F(2, 44) = 3.7314 [0.0319] All lags of LOAN_GROWTH F(2, 44) = 0.47178 [0.6270] All vars, lag 2 F(2, 44) = 0.17477 [0.8402]

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 124 10. Asset growth  NNIM

Equation 1: NNIM

Coefficient Std. Error t-ratio p-value const −0.00127924 0.000781822 −1.636 0.1083

NNIM_1 0.412098 0.131594 3.132 0.0030 ***

ASSET_GROWTH_1 −0.00292920 0.00782672 −0.3743 0.7099 time −2.94794e-05 1.76169e-05 −1.673 0.1008

Mean dependent var −0.003759 S.D. dependent var 0.001891 Sum squared resid 0.000130 S.E. of regression 0.001646 R-squared 0.286999 Adjusted R-squared 0.242436 F(3, 48) 6.440350 P-value(F) 0.000939 rho −0.004100 Durbin -Watson 2.000777

F-tests of zero restrictions: All lags of NNIM F(1, 48) = 9.8068 [0.0030] All lags of ASSET_GROWTH F(1, 48) = 0.14007 [0.7099]

11. Deposit growth  NNIM

Equation 1: NNIM

Coefficient Std. Error t-ratio p-value const −0.000699255 0.000875031 −0.7991 0.4284 NNIM_1 0.330295 0.155138 2.129 0.0388 * NNIM_2 0.147211 0.154500 0.9528 0.3458 DEPOSIT_GROW_1 −0.0136373 0.00660476 −2.065 0.0447 * DEPOSIT_GROW_2 0.00319913 0.00630251 0.5076 0.6142 time −2.85565e-05 1.91613e-05 −1.490 0.1431

Mean dependent var −0.003799 S.D. dependent var 0.001888 Sum squared resid 0.000117 S.E. of regression 0.001611 R-squared 0.344965 Adjusted R-squared 0.272183 F(5, 45) 4.739723 P-value(F) 0.001458 rho 0.044219 Durbin -Watson 1.883498

F-tests of zero restrictions: All lags of NNIM F(2, 45) = 4.6358 [0.0148] All lags of DEPOSIT_GROW F(2, 45) = 2.4223 [0.1002] All vars, lag 2 F(2, 45) = 0.58653 [0.5604] FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 125

12. Loan growth  NNIM

Equation 1: NNIM

Coefficient Std. Error t-ratio p-value const −0.00159148 0.000652761 −2.438 0.0185 *

NNIM_1 0.411009 0.131593 3.123 0.0030 **

LOAN_GROWTH_1 0.00160989 0.00442146 0.3641 0.7174 time −2.55597e-05 1.69799e-05 −1.505 0.1388

Mean dependent var −0.003759 S.D. dependent var 0.001891 Sum squared resid 0.000130 S.E. of regression 0.001646 R-squared 0.286888 Adjusted R-squared 0.242318 F(3, 48) 6.436857 P-value(F) 0.000942 rho 0.001042 Durbin -Watson 1.990814

F-tests of zero restrictions: All lags of NNIM F(1, 48) = 9.7552 [0.0030] All lags of LOAN_GROWTH F(1, 48) = 0.13258 [0.7174]

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 126 13. Asset growth  Risk

Equation 1: RISK

Coefficient Std. Error t-ratio p-value const −0.00458285 0.00429292 −1.068 0.2921 RISK_1 −1.02201 0.137593 −7.428 <0.0001 ** RISK_2 −0.735394 0.187340 −3.925 0.0003 ** RISK_3 −0.516632 0.146427 −3.528 0.0011 ** ASSET_GROWTH_1 0.0346492 0.0369000 0.9390 0.3534 ASSET_GROWTH_2 0.0234692 0.0390447 0.6011 0.5512 ASSET_GROWTH_3 −0.00114607 0.0381086 −0.03007 0.9762 time 7.06251e-05 8.50881e-05 0.8300 0.4115

Mean dependent var 0.000015 S.D. dependent var 0.010410 Sum squared resid 0.001769 S.E. of regression 0.006650 R-squared 0.652753 Adjusted R squared 0.591985 F(7, 40) 10.74168 P-value(F) 1.58e-07 rho −0.001495 Durbin-Watson 1.968732

F-tests of zero restrictions: All lags of RISK F(3, 40) = 23.295 [0.0000] All lags of ASSET_GROWTH F(3, 40) = 0.61712 [0.6080] All vars, lag 3 F(2, 40) = 7.3088 [0.0020]

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 127 14. Deposit growth  Risk

Equation 1: RISK

Coefficient Std. Error t-ratio p-value const 0.000220398 0.00393815 0.05596 0.9556 RISK_1 −1.00559 0.132334 −7.599 <0.0001 ** RISK_2 −0.738798 0.167586 −4.408 <0.0001 ** RISK_3 −0.559352 0.131420 −4.256 0.0001 ** DEPOSIT_GROW_1 −0.0238083 0.0285042 −0.8353 0.4085 DEPOSIT_GROW_2 0.0215756 0.0274601 0.7857 0.4367 DEPOSIT_GROW_3 −0.00485610 0.0274232 −0.1771 0.8603 time 5.54246e-06 7.90392e-05 0.07012 0.9444

Mean dependent var 0.000015 S.D. dependent var 0.010410 Sum squared resid 0.001790 S.E. of regression 0.006689 R-squared 0.648649 Adjusted R-squared 0.587162 F(7, 40) 10.54946 P-value(F) 1.97e-07 rho −0.017188 Durbin -Watson 2.016426 F-tests of zero restrictions: All lags of RISK F(3, 40) = 24.028 [0.0000] All lags of DEPOSIT_GROW F(3, 40) = 0.45417 [0.7158] All vars, lag 3 F(2, 40) = 9.0814 [0.0006]

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 128 15. Loan growth  Risk

Equation 1: RISK

Coefficient Std. Error t-ratio p-value const −0.00377320 0.00323298 −1.167 0.2501 RISK_1 −0.987600 0.142720 −6.920 <0.0001 ** RISK_2 −0.582968 0.189390 −3.078 0.0038 *** RISK_3 −0.427426 0.146318 −2.921 0.0057 *** LOAN_GROWTH_1 0.0552090 0.0195253 2.828 0.0073 *** LOAN_GROWTH_2 −0.0165639 0.0215850 −0.7674 0.4474 LOAN_GROWTH_3 0.00540699 0.0211289 0.2559 0.7993 time 6.44477e-05 7.44962e-05 0.8651 0.3921

Mean dependent var 0.000015 S.D. dependent var 0.010410

Sum squared resid 0.001540 S.E. of regression 0.006205

R-squared 0.697660 Adjusted R-squared 0.644750

F(7, 40) 13.18591 P-value(F) 1.16e-08 rho −0.023939 Durbin -Watson 2.005202

F-tests of zero restrictions: All lags of RISK F(3, 40) = 25.266 [0.0000] All lags of LOAN_GROWTH_d11 F(3, 40) = 2.6892 [0.0592] All vars, lag 3 F(2, 40) = 4.857 [0.0129]

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 129 16. Asset growth  Solvency

Equation 1: SOLVENCY

Coefficient Std. Error t-ratio p-value const 4.18423 1.00127 4.179 0.0001 ** SOLVENCY_1 0.452441 0.130593 3.465 0.0011 ** ASSET_GROWTH_1 0.538269 0.676570 0.7956 0.4302 time −0.00971286 0.00269215 −3.608 0.0007 **

Mean dependent var 7.209527 S.D. dependent var 0.321145 Sum squared resid 0.913577 S.E. of regression 0.137960 R-squared 0.826311 Adjusted R-squared 0.815455 F(3, 48) 76.11857 P-value(F) 2.91e-18 rho 0.036397 Durbin -Watson 1.918842

F-tests of zero restrictions: All lags of SOLVENCY F(1, 48) = 12.003 [0.0011] All lags of ASSET_GROWTH F(1, 48) = 0.63296 [0.4302]

17. Deposit growth  Solvency

Equation 1: SOLVENCY

Coefficient Std. Error t-ratio p-value const 5.03997 1.11211 4.532 <0.0001 ** SOLVENCY_1 0.451165 0.137148 3.290 0.0020 ** SOLVENCY_2 −0.122246 0.136716 −0.8942 0.3760 DEPOSIT_GROW_1 0.530636 0.491468 1.080 0.2860 DEPOSIT_GROW_2 1.47708 0.490759 3.010 0.0043 ** time −0.0110611 0.00292386 −3.783 0.0005 **

Mean dependent var 7.201987 S.D. dependent var 0.319658 Sum squared resid 0.757103 S.E. of regression 0.129709 R-squared 0.851812 Adjusted R-squared 0.835347 F(5, 45) 51.73376 P-value(F) 1.53e-17 rho 0.007515 Durbin-Watson 1.967536

F-tests of zero restrictions: All lags of SOLVENCY F(2, 45) = 5.5988 [0.0067] All lags of DEPOSIT_GROW F(2, 45) = 4.8094 [0.0128] All vars, lag 2 F(2, 45) = 4.6837 [0.0142]

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 130 18. Loan growth  Solvency

Equation 1: SOLVENCY

Coefficient Std. Error t-ratio p-value const 3.13218 1.40532 2.229 0.0317 * SOLVENCY_1 0.453430 0.160008 2.834 0.0072 **

SOLVENCY_2 0.193083 0.185116 1.043 0.3034 SOLVENCY_3 −0.143423 0.182466 −0.7860 0.4366 SOLVENCY_4 0.0834987 0.162041 0.5153 0.6093 LOAN_GROWTH_1 0.351570 0.429367 0.8188 0.4179 LOAN_GROWTH_2 −0.915978 0.439298 −2.085 0.0437 * LOAN_GROWTH_3 0.278957 0.472117 0.5909 0.5580 LOAN_GROWTH_4 0.948618 0.429418 2.209 0.0331 * time −0.00701017 0.00373811 −1.875 0.0682 *

Mean dependent var 7.184117 S.D. dependent var 0.313220 Sum squared resid 0.659526 S.E. of regression 0.130042 R-squared 0.859947 Adjusted R-squared 0.827627 F(9, 39) 26.60736 P-value(F) 5.62e-14 rho 0.055006 Durbin-Watson 1.860207 F-tests of zero restrictions: All lags of SOLVENCY F(4, 39) = 4.4201 [0.0048] All lags of LOAN_GROWTH F(4, 39) = 3.1823 [0.0235] All vars, lag 4 F(2, 39) = 3.5856 [0.0372]

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 131 Appendix 15. Impulse response analysis

1. Response of ROA in traditional banks to a shock in deposit growth in digital banks

response of ROA to a shock in DEPOSIT_GROWTH 0.0001

0

-0.0001

-0.0002

-0.0003

-0.0004

-0.0005

-0.0006

-0.0007

-0.0008 0 2 4 6 8 10 12 14 quarters

2. Response of ROE in traditional banks to a shock in deposit growth in digital banks

response of ROE to a shock in DEPOSIT_GROWTH 0.001

0

-0.001

-0.002

-0.003

-0.004

-0.005

-0.006

-0.007 0 2 4 6 8 10 12 14 quarters FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 132

3. Response of loans-to-assets in traditional banks to a shock in loan growth in digital

banks

response of RISK to a shock in LOAN_GROWTH 0.003

0.002

0.001

0

-0.001

-0.002

-0.003 0 2 4 6 8 10 12 14 quarters

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 133 4. Response of liabilities-to-equity in traditional banks to a shock in deposit growth in

digital banks

response of SOLVENCY to a shock in DEPOSIT_GROWTH 0.06

0.05

0.04

0.03

0.02

0.01

0

-0.01 0 2 4 6 8 10 12 14 quarters

5. Response of liabilities-to-equity in traditional banks to a shock in loan growth in digital banks

response of SOLVENCY to a shock in LOAN_GROWTH 0.04

0.03

0.02

0.01

0

-0.01

-0.02 0 2 4 6 8 10 12 14 quarters FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 134 Appendix 16. Results of White’s test for heteroscedasticity

# of model Model p-value Significance OLS_1 Deposit growth-->ROA 0.004943 *** OLS_2 Deposit growth-->ROE 0.004861 *** OLS_3 Loan growth-->RISK 0.004702 *** OLS_4 Deposit growth-->SOLVENCY 0.000595 *** OLS_5 Loan growth-->SOLVENCY 0.01502 ***

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 135 Appendix 17. Full test results for OLS analysis

1. Deposit growth  ROA

Model 14: Heteroskedasticity-corrected, using observations 2005:4-2017:4 (T = 49) Dependent variable: ROA

Coefficient Std. Error t-ratio p-value const 0.00342052 0.000355017 9.635 <0.0001 *** DEPOSIT_GROWT_1 7.90966e-05 0.00418116 0.01892 0.9850 DEPOSIT_GROWTH_2 −0.00437310 0.00250693 −1.744 0.0884 * DEPOSIT_GROWTH_3 −0.00953934 0.00396905 −2.403 0.0207 ** DEPOSIT_GROWTH_4 0.00158599 0.00337779 0.4695 0.6411 FEDFUNDS_RATE 0.119117 0.0636367 1.872 0.0682 * CRISIS −0.000967887 0.000466930 −2.073 0.0444 **

Statistics based on the weighted data: Sum squared resid 124.4769 S.E. of regression 1.721551 R-squared 0.563204 Adjusted R-squared 0.500804 F(6, 42) 9.025783 P-value(F) 2.40e-06 Log-likelihood −92.36934 Akaike criterion 198.7387 Schwarz criterion 211.9814 Hannan-Quinn 203.7630 rho 0.505173 Durbin-Watson 0.989652

Statistics based on the original data: Mean dependent var 0.002553 S.D. dependent var 0.001638 Sum squared resid 0.000077 S.E. of regression 0.001353

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 136 2. Deposit growth  ROE

Model 16: Heteroskedasticity-corrected, using observations 2005:4-2017:4 (T = 49) Dependent variable: ROE

Coefficient Std. Error t-ratio p-value const 0.0246527 0.00288969 8.531 <0.0001 *** DEPOSIT_GROWTH_1 0.0150767 0.0337674 0.4465 0.6575 DEPOSIT_GROWTH_2 −0.0127538 0.0282057 −0.4522 0.6535 DEPOSIT_GROWTH_3 −0.0568608 0.0337792 −1.683 0.0997 * DEPOSIT_GROWTH_4 0.0116093 0.0283238 0.4099 0.6840 FEDFUNDS_RATE 0.951963 0.624427 1.525 0.1349 CRISIS −0.00800081 0.00342967 −2.333 0.0245 **

Statistics based on the weighted data: Sum squared resid 96.83115 S.E. of regression 1.518388 R-squared 0.365867 Adjusted R-squared 0.275277 F(6, 42) 4.038701 P-value(F) 0.002777 Log-likelihood −86.21612 Akaike criterion 186.4322 Schwarz criterion 199.6750 Hannan-Quinn 191.4565 rho 0.514599 Durbin-Watson 0.970545

Statistics based on the original data: Mean dependent var 0.020394 S.D. dependent var 0.013743 Sum squared resid 0.005851 S.E. of regression 0.011803

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 137 3. Loan growth  RISK

Model 20: Heteroskedasticity-corrected, using observations 2005:3-2017:4 (T = 50) Dependent variable: RISK

Coefficient Std. Error t-ratio p-value const −0.00149970 0.00171803 −0.8729 0.3874 LOAN_GROWTH_1 0.0549260 0.0295557 1.858 0.0698 * LOAN_GROWTH_2 −0.0357887 0.0131866 −2.714 0.0095 *** LOAN_GROWTH_3 0.0188642 0.0162674 1.160 0.2525 FEDFUNDS_RATE 0.0415059 0.510218 0.08135 0.9355 CRISIS −0.00134319 0.00445205 −0.3017 0.7643

Statistics based on the weighted data: Sum squared resid 115.8111 S.E. of regression 1.622366 R-squared 0.304770 Adjusted R-squared 0.225766 F(5, 44) 3.857673 P-value(F) 0.005515 Log-likelihood −91.94536 Akaike criterion 195.8907 Schwarz criterion 207.3629 Hannan-Quinn 200.2594 rho −0.714179 Durbin-Watson 3.422426

Statistics based on the original data: Mean dependent var −0.000072 S.D. dependent var 0.010206 Sum squared resid 0.004571 S.E. of regression 0.010192

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 138 4. Deposit growth  SOLVENCY

Model 22: Heteroskedasticity-corrected, using observations 2005:2-2017:4 (T = 51) Dependent variable: SOLVENCY

Coefficient Std. Error t-ratio p-value const 6.83129 0.0546063 125.1 <0.0001 *** DEPOSIT_GROWTH_1 2.34557 0.669629 3.503 0.0010 *** DEPOSIT_GROWTH_2 2.69851 0.687547 3.925 0.0003 *** FEDFUNDS_RATE −9.25084 14.5516 −0.6357 0.5281 CRISIS 0.172915 0.0699901 2.471 0.0173 **

Statistics based on the weighted data: Sum squared resid 71.56051 S.E. of regression 1.247262 R-squared 0.676077 Adjusted R-squared 0.647909 F(4, 46) 24.00223 P-value(F) 9.10e-11 Log-likelihood −81.00317 Akaike criterion 172.0063 Schwarz criterion 181.6655 Hannan-Quinn 175.6974 rho 0.749597 Durbin-Watson 0.381490

Statistics based on the original data: Mean dependent var 7.201987 S.D. dependent var 0.319658 Sum squared resid 3.813755 S.E. of regression 0.287937

FINANCIAL INNOVATION IMPACT ON TRADITIONAL FINANCIAL INTERMEDIARIES 139 5. Loan growth  SOLVENCY

Model 24: Heteroskedasticity-corrected, using observations 2005:4-2017:4 (T = 49) Dependent variable: SOLVENCY

Coefficient Std. Error t-ratio p-value const 6.86064 0.0464936 147.6 <0.0001 *** LOAN_GROWTH_1 1.86799 0.555998 3.360 0.0017 *** LOAN_GROWTH_2 −0.0385897 0.497804 −0.07752 0.9386 LOAN_GROWTH_3 0.767854 0.718280 1.069 0.2912 LOAN_GROWTH_4 2.14455 0.600787 3.570 0.0009 *** FEDFUNDS_RATE −14.8570 7.58373 −1.959 0.0568 * CRISIS 0.289808 0.0706773 4.100 0.0002 ***

Statistics based on the weighted data: Sum squared resid 63.57435 S.E. of regression 1.230315 R-squared 0.796793 Adjusted R-squared 0.767764 F(6, 42) 27.44768 P-value(F) 4.81e-13 Log-likelihood −75.90754 Akaike criterion 165.8151 Schwarz criterion 179.0578 Hannan-Quinn 170.8394 rho 0.633701 Durbin-Watson 0.557727

Statistics based on the original data: Mean dependent var 7.184117 S.D. dependent var 0.313220 Sum squared resid 3.039491 S.E. of regression 0.269015