The influence of regulatory and supervisory practices on worldwide adaptation to financial innovation

(1)

Master Thesis

The Influence of Regulatory and Supervisory Practices on

Worldwide Adaptation to Financial Innovation

Author: J.E. van Rhee

Student Number: 10640738

Date of Submission Final Version: August 31st_{, 2015}

Program: MSc Business Administration

Track: Innovation and Entrepreneurship

Institution: Amsterdam Business School, University of Amsterdam

Supervisor: Drs. A.C.C. Gruijters

(2)

Statement of originality

“This document is written by student Jurjen Elmer van Rhee who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.”

(3)

Abstract

Regulatory aspects have influence on the adaptation to financial innovation for most countries worldwide. The aspects to which it reacts however differs among continents and especially regulatory types. This analysis is possible by creating a new way to compare regulatory systems worldwide. This system is based on the quantification of the World Bank Regulatory Survey. This resulted in the first comparable assessment of worldwide regulatory systems, which was presented in a regulatory matrix with Power (Ability) and Stringency (Strictness) as axis values. For further analysis, it is linked to the Global Financial Development Database from the International Monetary Fund. The performed analysis regressed eight variables and a control variable for a set of 120 countries and various subsets based on Gross National Income per Capita, Regulatory type and Geographic location. The results show that the adaptation of Financial Innovation does indeed link to regulatory aspects. This affect does differ among continents (with strong effects in Africa) and regulatory systems (strong effects for Strict / Able countries).

(6)

Introduction

This thesis empirically tests the relationship between regulatory and supervisory aspects in the financial sector and the adaptation of financial innovation. It finds strong relations for industry structure, regulatory strictness and Gross National Income per Capita. These are

especially apparentfor Africa, which was not been researched before and for countries with

powerful and strict regulators. It also creates a much needed framework to compare financial regulatory and supervisory systems worldwide.

Financial innovation has influenced society since the existence of money. It has gotten us all the advantages that come with a modern economy. The ability to use an ATM, wire money, invest in firms and take out mortgages are all results of financial innovation that people use daily. Most of these services are a result of financial intermediation (the brokerage function) or asset transformation, the core tasks of banks and other financial intermediaries. Banks reduce the information asymmetries in the market and therefore create a more efficient market, allowing all players in it to benefit (Greenbaum & Thakor, 2007).

The result of financial innovation is not just purely positive. The 2008 stock market crash however, is also attributed to financial innovation. The complexity involved with current stock markets filled with high frequency traders, complex derivatives such as Credit Default Swaps and products such as synthetic Collateralized Debt Obligation. They have brought efficiency to the market by making home mortgages tradable, but also greatly increased the information asymmetries in it by complicating the ownership structure (Crotty, 2009).

The benefits of financial innovation have only reached part of the world. 2.5 billion adults globally are still not included in the financial system, of which 2.2 billion (62% of total

(7)

adults) are located in Africa, Asia, Latin America and the Middle East (Chaia et al., 2013). Anyone would notice that these adults are not located in ‘high income economies’. They are located in ‘upper-middle-income economies’ or ‘lower-middle-income’ economies (World Bank definitions). The correlation between income and access to financial services seems apparent. The causal relation however is more complex. It would make sense that with higher banking efficiencies the economy would be more likely to advance to a ‘high income economy’. The increased efficiency of these financial innovations lowers the costs of transactions and the availability of investments, prospering the economy. This is the train of thought building on Schumpeter (Schumpeter, 1934). On the other hand, those same high income economies would develop a higher demand for financial services due to higher assets and disposable income. People would want to invest their money and demand more opportunities and service. This would then also lead to the adaptation of financial innovation. This would follow the likes of Robinson (Robinson, 1953). The correlation is confirmed by Allen & Ndikumana (2000), who performed the research on a set of sub-Saharan countries with different growth patterns, and others (Beck & Demirguc-Kunt, 2006; Beck & Demirgüç-Kunt, 2008).

To check this research for today’s applicability a simple analysis is performed using publically available data on Financial Development and the country’s GDP per capita (as done by many studies including Knack & Keefer (1995)). Financial development is measure by the World Economic Forum (2012) who score the financial development of 62 countries on seven pillars. Once gathered, the data plotted and tested shows a statistically significant (at α = 0.01) relation, explaining 61.6% of the variance. There is an apparent strong correlation

(8)

reason that countries with vastly different GDP per capita numbers have the same level of financial development.

Figure 1 shows on the vertical axis the average GDP per capita in the year 2012. The horizontal axis shows the score on the Financial Development Index. There is a statistically significant relation between the two in this sample. Source GDP: International Monetary Fund. Source Financial Development: World Economic Forum (2012).

Causal evidence on this GDP / Financial development relationship is provided by various authors (Demirgüç-Kunt & Klapper, 2012; King & Levine, 1993; Levine, 1997; Levine, 1998; Levine, 2005). Most of these papers mainly researched developed countries because of data availability. This was the reason why Allen & Nikumana (2000) only used a sample of 7 countries. Alongside those, analysis by Chaia et al. (2013) found that the levels of financial inclusion are influenced by innovation, independent of countries socio-economic and

(9)

demographic characteristics. They are more specifically influenced by the adoption of innovation. Bank accounts and loans have existed for centuries, but have yet to reach a large part of the world. Chaia also note a lack of available data.

The relationship between financial innovation and economic prosperity is apparent. Laeven, Levine, & Michalopoulos, (2015) determine that although technological advancement has helped financial advancement and economic growth, innovation is needed to further economic growth. Financial innovation plays an important role in the advancement of countries and could therefore has a large influence on the 2.5 billion people who are not yet included in the financial system.

Since it is impossible to just ‘switch on’ innovation in those countries, it is necessary to first examine the most important determinants of financial innovation. If the goal is to let prosper low-income economies, it is apparently possible to speed up that process if financial innovation is advanced in those economies. What makes certain countries better or worse adaptors of financial innovation? Although there are multiple factors, the regulatory aspect is chosen to focus on.

The Financial Services industry is one of the most heavily regulated industries worldwide. Virtually all countries have central banks that grant bank licenses and oversee the stability of the banking system. Many other countries are part of a larger banking system, such as the European Central Bank (ECB). Although the efficiency and effectiveness of this regulatory system is at least debatable (Barth, Caprio, & Levine, 2004; Bhattacharya, Boot, & Thakor, 1998), its existence is a given. All of the studies mentioned earlier, for example (Chaia et al., 2013), point at regulatory practices as a factor in bank innovativeness, bank performance and therefore economic performance. A strongly performing regulatory system ensures stability in the economy and therefore lower information asymmetries.

(10)

These banking systems are different throughout the world. They vary from very sophisticated to very rudimental, which therefore influences economic development of the country involved (Barth, Caprio Jr, & Levine, 2001; Dev, 2006). There has been a lot of research in the field regarding the influence of regulation on banking. Much of that is related to the USA and some is related to Europe. However, there is no generally applicable research that could help developing countries with their regulatory system. Their economies are growing and dependent on the financial inclusion of its people.

There is however research on the influence of Regulation on Innovation that is generally applicable. There are roughly three types of regulations that cause innovations. (Blind, 2012) • Regulation that is focused on promoting innovation - intellectual property rights.

• Regulation that influences a company’s strategy and activities, but not innovation in a positive sense. This results in a regulatory burden and compliance costs.

• Other objectives: Protecting health and safety. This will lead to innovation to adapt to or circumvent regulation

All of these are controlled by regulators. There is no literature disputing the influence of regulation on financial innovation. It has a strong influence on innovation in the financial sector and has therefore influenced the course of the development. However, it can also be a hindrance (Frame & White, 2014).

It is unclear what type of regulation is most favorable to (adaptation of) financial innovation. Other research (including the World Economic Forum) mainly uses macro data to establish the type and extend of financial development. This thesis will use the Global Financial Development Database, also known as Findex (Demirgüç-Kunt & Klapper, 2012). This

(11)

enables me to measure financial innovation by the results in the real economy. This database is extensive and offers 148 countries in all income classes.

Until recently there was a lack of possibilities to establish the type of regulation in countries, especially in developing countries. Recently the World Bank has conducted a very extensive Regulation Survey with respondents from 142 central banks. There was an earlier, less extensive version of this survey. The last (World Bank, 2011) version covers 630 banking regulatory and supervisory items for 142 countries. This gives a clear insight in the type of regulatory power and activities in the vast majority of countries worldwide. Combining databases on Financial Development and the Regulatory Survey has not been done before, at least not in published form. This large cross-section gives new insights on the influence of regulation in financial innovation and inclusiveness. This research idea is aided by the research suggestions from Frame and White (2004).

The research question in this thesis is:

How does the type and strength of financial regulatory practices influence the adaptation of financial innovation?

`

Sub questions:

1. How do financial regulatory systems compare worldwide?

2. What characteristic of the regulatory system has the most influence? 3. What are the differences of this regulatory influence among continents?

4. What is the difference in this relationship between high-, middle- and low income economies?

5. Which aspects of financial regulatory practices work best in strict or not strict systems?

(12)

This thesis contributes to existing literature in four ways. Firstly, is the first research that empirically tests the relationship between types of financial regulation and financial innovation. Secondly, the first to empirically test several types of regulatory practices and their relationship to financial innovation and the adaptation of innovation. Thirdly it is the first to expand the horizon of this type of research beyond OECD countries, by empirically testing for 120 countries worldwide. And lastly it is the first to characterize regulatory systems by power and strictness, allowing to group regulatory systems on the extend of their influence, not specific laws.

In its design the thesis combines law creation by governments, business and economic indicators and is therefore by nature multidisciplinary. It uses regulations set by governments to regulate the financial industry, which influence the (usually) private firms operating in it, and measures the success on the adaptation of financial innovation measured by indicators from the real economy.

This thesis will have the following outline. It will first reflect on the existing literature. This will focus on the research on the general connection between regulation and innovation, from the financial and other sectors. This will give a broader perspective on how the two connect. Then it will explain the different types of financial regulatory systems and their possible effects on the economy and innovation. This gives the necessary background to understand the differences among countries. Main subject in the literature review is the relationship between regulation and financial innovation. This will provide the hypotheses that will be tested. Secondly, the methodology, data collection and transformation will be explained. Lastly, in results and conclusions the research questions and will be answered and extended with further research suggestions and limitations.

(13)

Literature Review

The literature review covers the following parts and will cover all general topics related to the influence of regulation on financial innovation. It will cover three general topics before addressing the existing literature on the research question. This framing is necessary since the research builds on concepts built in the general field. Below is a schematic representation of structure of the literature study.

Definitions on innovation

The original research on innovation describes innovation as the process of structural change, described in five types (Blind, 2012):

1. The launch of a new product or a new species of already known product.

2. An application of new methods of production or sales (not yet proven in the industry) of a product.

3. The opening of a new market (the market for which a branch of the industry was not yet represented)

4. The acquiring of new sources of supply of raw material or semi-finished goods. 5. The new industry structure such as the creation or destruction of a monopoly position.

Figure 2 Shows the interrelatedness between the literature. The final research question pulls literature from various subfields, which are all listed

(14)

Economic growth is fuelled by new combinations of resources made by entrepreneurs, rather than independent forces. With an incline in economic capital, more resources will be available for research and development which will result in innovation. This is worthwhile since the innovator may endure Schumpeterian rents, which come to the innovator (Schumpeter, 1934). Note that Schumpeter includes not just inventions, but also new markets (with existing products) and puts commercial success as a condition for innovation.

This is in stark contrast to neoclassical economists like Solow. In his view growth only comes from capital investments, labor force growth and depreciation rate. Growth is only available through technological progress (inventions). Sustained growth is therefore limited to the developments in technology (Solow, 1956).

Both the Schumpeterian (Stewart, 2010) and the Solow definitions of innovation (Carlin & Soskice, 2005) are still used in modern economic literature although some choose to combine them (Crafts, 2006). This thesis will use the definition of innovation as postulated by Schumpeter, therefore including opening of new markets and new industry structures. For the avoidance of misinterpretation this will be denoted as adaptation to innovation.

A further important aspect to innovation is the effect of competition. In the Schumpeterian world, there should be an endless benefit to innovation. Firms that innovate should always receive the benefits of innovation in the form of Schumpeterian rents. However, the time it takes the competition to catch up is important in the duration of those rents. Other literature shows that this might not be the case Competition indeed seems to have a positive effect on innovation, but the relationship takes the form of an inverted U-Shape. Meaning that when competition gets heavier, eventually firms will stop innovating (Amable, Demmou, & Ledezma, 2009). This is caused by the larger profits from imitating other innovations, rather than developing their own (Cornaggia, Mao, Tian, & Wolfe, 2015; Teece,

(15)

1986). This is contrary to Porters believe on competition that competition forces firms to innovate in order to stay in the game (Porter, 1990).

Innovation in financial services

Banks usually innovate in services and non-physical products, and never in physical products. And the products banks are selling (swaps, mortgages etc.) are usually a way for the bank to deliver a service. A swap gives the client a different exposure to interest rates and a mortgage gives a client the ability to purchase a house with external financing. Therefore, most banks are seen as service providers.

It is important to note that the banking industry exists of two specific parts, Investment Banking and Retail banking. The former underwrite IPO’s, securitize mortgages and create CDO’s. The only deal with institutional investors, large corporations and other institutions (Ritter, 2003). Retail banks are the kind that the general public hold accounts with. They transfer money, sell mortgages and savings accounts. In Europe these can be two sides of the same bank, where in the USA the combination of the two is illegal (Turner, 2009). They are largely subject to different regulations and supervisors, which also differs among countries. However, the innovation process is subject to the same mechanisms. Therefore they can be analyzed together (Levesque & McDougall, 1996).

The ability to use an ATM, wire money, invest in firms and take out mortgages are all results of financial innovation that people use daily. The ability to transfer and keep money in a safe way would be an innovation for countries that do not yet have access to these aspects. For developed countries, innovation in financial services focusses on online and mobile banking.

(16)

Bank Innovation

All cited papers on bank innovation mention regulation as a source of innovation within banks. This is either as a direct cause of innovation (Barras, 1990; Cornaggia et al., 2015; Silber, 1983), an obstacle for innovation (Ahn, 2002; Silber, 1983).

Banking as a sector has gone through major changes since the emergence of computers. From the 1980’s on, banks were very suitable for rapid innovation with the invention of more advanced computer systems. The financial service sector was in “the right condition for a rapid initial rate of take up, (…) leading to spectacular growth and the most far-reaching set of process and product innovations”. The industry was so favorable for this innovation since it is “basically unstructured and with little political regulation” (Barras, 1990). In this time frame, all innovation was due to exogenous factors as inflation, tax effects, legislation or internalization (Silber, 1983).

All these exogenous activities had a large impact on the banking sector. The sector changed from closed and slow to very open and dynamic. This opened the market up to competition, which in turn increased the need for innovation (Ahn, 2002; Vermeulen, 2004). Banks who became subject to more competition actually showed less innovative activities. On the other hand, banks who were not subjected to competition but to additional markets (i.e. became the competitor) were more prone to innovate. This was tested in a sample of deregulated banks, who became subject to competition (Cornaggia et al., 2015).

The difference probably originates from the method and definition of ‘innovation’. The qualitative research resulted in the expected effect, where the quantitative research resulted in an opposing view. Cornaggia measured innovation in patents. Since banking is a service industry, patents is a difficult measure to grasp the other four parts of the Schumpeterian definition of innovation. This might explain the difference.

(17)

Less is known about how banks innovate internally. There are four large barriers to innovation in banks (Vermeulen, 2004): The biggest challenges are firstly the departmentalized structure of banks, which leads to a battle of resources and communicational problems. Secondly the limited use of New Product Development Tools, thirdly the very conservative and risk avoiding culture. Lastly, banks have problems integrating their IT systems and innovating on that. This is due to the fact that most banks have outsourced their IT-development to try and accelerate their innovation process (Drew, 1995). The banks innovation is also diffused by stability (Vermeulen, 2004), which is the contrary goal of the regulators (Tarullo, 2012). Banks in stable environments are not incentivized to innovate due to the absence of competition and regulatory compliance (Vermeulen, 2004). This puts regulators in a though spot since they prefer innovation in banking and a stable environment.

There is a contrasting theory of innovations in the banking industry. It proposes a reversed product cycle for innovations in services, and particularly financial, industries. Usually, the innovation would start with a market demand, let’s say easier access to checking accounts. Normally this would be a market gap that companies would try to fill by innovating in their products or technologies. There is however evidence that this is actually reversed for (financial) services firms.

Barras (1990) mentions three stages of financial service innovation. First is an investment in new technology. This investment is usually made by large incumbent firms, for who this technology is profitable. The first goal is usually efficiency. The second stage is more radical, using the now existing technology for improved effectiveness. This improves the quality of service to existing customers. The third stage moves more towards product innovations, based on new technological advancements. This will have overall impact on the

(18)

firms output (ATM’s in the case of the example mentioned before) resulting in economies of scale and growing capital savings.

Regulation and innovation

An important aspect of every company is the environment it finds itself in. No firms are exempted from the environment in which they operate. A large aspect of that environment is the public sector, the law-creating part of the economy. The public sector for example sets taxes, minimum wages, grant patents and sets carbon exhaust limits. This directly influences the actions and the profitability of the companies who are present in their jurisdiction.

In the literature there is consensus on the types of regulations available. Lundvall (2010) describes two types of regulations. The first aiming at economic efficiency (e.g. Patents) and the second which pursues other goals (e.g. Environmental). Stewart (2010)

refers to these as ‘Social’ regulations. Blind (2012) adds a third: institutional.1_The

classification however differs vastly among authors. Lundvall describes patents as an economic regulation, whereas Blind calls them institutional. Another paper refers to them as ‘information’ regulation (White, 2000). This thesis follows the definitions used by the (OECD, 1997) Effects in italics are from Blind (2012). They are as follows:

Types of regulations: -‐ Economic

o Intervene directly into market decisions such as pricing, competition, market entry or exit.

o Aim at improving economic efficiency. o Usually positive effects on innovation.

1_{Blind (2012) claims to be following the OECD (1997), but they use ‘Administrative}

(19)

-‐ Social

o Protection of individuals and environment on topics such as health, safety, the environment and social cohesion.

o Explicit policy instrument.

o Economic concerns might be secondary.

o Positive effects due to compliance efforts.(Porter & Van der Linde, 1995) -‐ Administrative

o Paperwork and administrative formalities. This is where governments collect information and intervene in individual economic decisions.

o Can have substantial impact on the economic sector.

o Include Patent laws, bankruptcy protection and employment regulation.

o Positive effects in the case of patents, property rights and loosened bankruptcy

regulations (ambiguous results).

o Negative effects for administrative burdens and tighter bankruptcy regulations

(ambiguous results).

There are two main effects for companies regarding regulation compliance, and incentive effects (Carlin & Soskice, 2005). The reaction to regulation differs wildly among sectors and companies. However, there are many common features to the responses.

If regulation is able to change the incentives for investing, it is likely to affect innovation. This has a short term and long term effect. First compliance costs influence investments. If the government or regulatory institutions impose new regulations, it takes companies at the very least time and effort to comply with those regulations. This time could have been used on a productive task, therefore imposing a cost on the company. Compliance costs are found to have negative short term influence on innovation. This happens because the compliance costs have a tax-like burden on the capital available. This takes its strain on R&D. It also lowers capital intensity which, following the Schumpeter relation, would lower innovation.

After the initial compliance costs, the incentive impact follows. This is often, depending on the type of regulation, a positive long run effect on innovation. Older according to Blind (2012) is much more ambivalent to the positive effects of regulation on innovation

(20)

than more recent literature, of which the majority is positive (Blind, 2012) The strongest effect is witnessed in lightly regulated countries that were heavily regulated before i.e. liberalized.

The incentive effects of regulation take one of two forms. It either leads to circumventive innovation or compliance innovation. Circumventive innovation tries to get around the new regulation. This occurs when the regulation has a narrow scope and can easily be evaded by slightly changing the product or service. Compliance innovation is the result of broader regulation, where slight changes to the product or service are not able to circumvent it. Note that this process will not result necessary into an innovation in the Schumpeterian sense of the word.

Compliance innovation does not per definition equal commercial success. Still, this is where most of the literature is focused. In the end, if the results from compliance innovation exceeds the compliance costs there is a net effect of regulation through innovation. In the light of advancing the economy, it would therefore be most positive to issue regulation (to trigger compliance innovation) with low compliance cost.

This is easier for larger companies, who are able to afford the compliance costs (Blind, 2012; Stewart, 2010). There is even a first-mover advantage to be had. If firms are regulated earlier than their competitors, they incur a loss at first. Eventually they will be able to capture market share from their competitors who are regulated after them (Porter & Van der Linde, 1995). It must be noted here that all available research has been performed in the USA or OECD countries, which are primarily high income economies. Conditions in low income economies might be vastly different, due to different market structures and the spreading of technology.

(21)

Types of regulations and regulatory institutions in the financial services industry.

There are usually two parts in a bank regulatory system: regulation on structure and conduct (Freixas & Rochet, 2008; Greenbaum & Thakor, 2007). The first is the regulation of structure (prudential supervision). This controls the bank licenses and decides which company is allowed to offer certain cervices. It is mainly concerned with systemic risk and the stability of the financial system as a whole. This task is usually carried out by a central bank of some sort. Examples are the FED in the USA, or the De Nederlandse Bank (DNB) in the Netherlands. For Europe, the European Central Bank (ECB) supersedes the national central banks and holds direct control over the large banks form member states (European Central Bank, 2014).

The second is regulation of conduct. It regulates the behavior of individual banks and establishes what they are allowed to do within their license. An example for the USA would be the Securities and Exchange Commission (SEC) or the Autoriteit Financiële Markten (AFM) in the Netherlands. These are governmental bodies that research the sector, advise the government on laws and have mandate to issue fines.

Together they hold six types of regulatory instruments (Freixas & Rochet, 2008) 1. Deposit interest rate ceilings

2. Entry, branching, network, and merger restrictions 3. Portfolio restrictions, which include reserve requirements 4. Deposit insurance

5. Capital requirements

6. Regulatory monitoring and supervision.

Although this is a very broad overview of the types of regulations available, they are very common throughout the world. And although regulations have become more and more

(22)

similar there is still a wide diversity in the type and the manner of regulatory practices per country (Barth, Caprio, & Levine, 2004).

Regulation and Financial Innovation

Since the financial services are among the most heavily regulated sectors in the world, it could be expected the relationship between regulation and innovation is strong in this field. Financial innovation is often viewed as a byproduct of regulation (Silber, 1983). All papers on innovation in the financial services industry mention regulation as one of, or the main determinant of innovation (e.g. Berger, Kashyap, Scalise, Gertler, & Friedman, 1995; Merton, 1995; Miller, 1986; White, 2000). There are three points of criticism; the literature is limited to a handful of countries, no new contributions have been added in the last ten years and there is no shared vision on the optimum of regulation for the sector. However, since all available literature focuses on OECD countries, there is no way to compare regulatory systems worldwide, let alone rank them.

Most papers on the topic are relatively old. Written before the 2008 financial crisis and the internet bubble, they do no encompass any of the recent changes in finance. They missed the implementation of the Basel accords, securitization and mobile banking. There is one recent paper on the relationship which ties regulation and innovation together for the past 30 years (Frame & White, 2014). It builds, however, on a decade old paper by the same authors (Frame & White, 2004). There is also another paper, but no empirical testing of the relationship has been done since 2005 (Frame & White, 2004).

There is no literature on the optimum of bank regulation, implying that there is no theory on the exact relationship. This should be possible when broad data on regularity structures becomes available (Jackson, 2007).

(23)

In theory, innovation in banking should not respond differently than in other sectors. Banks are still subject to compliance costs and regulatory incentives. An overview of the available literature states that this is the case (Frame & White, 2004). Financial regulation has had its influence on innovation in this sector, and vice versa (White, 2000). However where most research focusses on compliance innovation the financial sector’s innovation is more than in other sectors driven by circumventive innovation (Silber, 1983). This could be a response to heightened compliance costs by over-specific regulation. This gives banks the incentives and the possibilities to evade the regulations (Warren, 2008). Prior, older empirical tests all conclude that regulation does indeed greatly influence innovation in the financial sector (Baer & Pavel, 1988).

Conclusion

There is limited research on the influence of regulation on the adaptation to innovation in the financial sector, or to innovation in the financial sector itself. This is especially true for countries outside of the first world. It is clear that regulation has strong influence on the financial sector as a whole, and on innovation inside the sector. It is also clear that regulators are able to allow or deny new entrants and products to the market. Regulatory influence is an important factor in banks’ actions, and therefore in the financial inclusion of the customer.

(24)

Hypotheses:

This leads to the following hypotheses:

H1 There are differences in results between North-America & Europe and developing

continents like Africa.

H2 Countries with lower income will show greater response to regulation i.e. have

more significant variables.

H3 Different regulatory groups2 will show different relationships with the dependent

variable.

H4 Strict and Able regulators will have more significant regulatory aspects than Not

Strict and Not Able regulators2.

H5 Regulatory influence will be stronger in countries with low adoption of financial

innovation than in countries with already high levels of financial innovation.

H6 GNI per capita is has a positive relationship with the adaptation to financial

innovation.

Some hypotheses are not directly derived from the literature, when there is not literature available on the subject (2 and 5) It is difficult to say anything regarding the direction of these relations. The sample does not just include high income countries. This make comparing to pervious literature not possible On the basis of literature on general regulation and innovation, an inverted U-Shaped graph is expected (similar to (Amable et al., 2009)) when plotting GNI/capita figures and the innovation effects of regulation. In a country with no regulation, there is a low incentive for banks to claim the market. It leaves them vulnerable for all changes to the market. If, for example, debt contracts are not formalized with an enforceable notary system the risks of lending to inhabitants is very high. For these countries stricter regulation would probably be beneficiary. On the other end of the spectrum in the high income countries, more regulation would impose more compliance costs. There is little to be gained in this saturated market. The extra costs would be taken out of the budget for R&D or other improvements, but would certainly not benefit innovation.

2_{The groups are defined using the analysis of regulatory groups and the regulatory matrix in}

(25)

Methodology

The methodology of this thesis has two goals. First the development of a method for comparison of regulatory systems worldwide. It is seeking to create a working division of worldwide regulatory systems, which is not available at the moment since they are difficult to compare. The Barth (2013) analysis however opens up possibilities to do just that. The second goal is to creating a method to test the relationship between regulation and financial innovation worldwide.

The methodology builds on two main databases (World Bank Regulatory Survey & Global Financial Development Database) and two main papers (Barth, Caprio Jr, & Levine, 2013; Demirgüç-Kunt & Klapper, 2012). An indicator of financial innovation per country will be regressed with a group of regulatory and supervisory attributes with Gross National Income per Capita as a controlling variable. This will answer which type of regulations and supervisory practices influence financial sector innovation.

Figure 3 is shows the researched relation between the independent and dependent variables. This is also the set up of the regressions. The data for the independent variables comes from The World Bank Regulatory Survey by the World Bank, whereas the data for the Financial Sector Innovation Indicator comes from the Global Financial Development Database.

(26)

A regression on the main sample is only able to show general trends. The real applicable results will be visible in the more detailed regressions. The results for variables might be different in groups of countries. Imagine regulators in a low-income low-tech economy as Bangladesh or a high-income high-tech economy as the Netherlands. Tighter regulations might have vastly different results in both countries. This is the reason why the partial regressions are important. They will show the effect of each specific part of regulation among similar countries.

Countries will therefore be grouped by GNI per capita, since this largely correlates with financial advancement. The low dispersion in GNI per capita enables the research to show differences in regulatory practices. The dataset is split into 5 groups, each containing 20% of the countries the dataset. Starting from the lowest 20% of GNI per capita to the highest. This is tested on a 2011 cross section of all available countries. Since regulations per country have a very low variation over time, it is unnecessary to perform panel data analysis. The time series component therefore does not add much to the regression and is therefore not used, leaving the cross-sectional regression using the 2011 data.

Data

The data used in this thesis consist of databases from the World Bank and the International Monetary Fund (IMF), which have not been linked in research previously. The World Bank provides data on per country supervision though their unique World Bank Regulation Survey. The World Bank surveys central banks in the entire world on their regulatory practices. Both response rate and number of questions have increased over time, reaching 142 and 630 respectively in 2011. The survey occurs every four years and is improved each time by input from officials and academics. This gives enormous amounts of input to classify and rank all 142 countries for regulatory practices. Barth et al. (2013) have developed a method of quantifying the questionnaire. They however stop at the quantification of the questions. By

(27)

extending their research into testable variables, it is possible to see the influence of these variables on the financial development of the regulated country. From the IMF this thesis uses the Global Financial Development Database (Findex). It offers extensive data for 203 economies. It includes per country data on:

Financial depth The size of financial institutions and markets

Access Degree to which individuals can and do use financial services

Efficiency Efficiency of financial intermediaries and markets

Stability Stability of financial institutions and markets

The database is built on solid academic research, most of the authors who worked on it have been cited in this thesis (Cihak, Demirgüç-Kunt, Feyen, & Levine, 2012). This thesis combines indicators to create one number for Financial Development with characteristic that measure adaptation to financial innovation. It does this by using key indicators from the dataset and transforming them into a percentage of the worldwide maximum for that value. This makes the values comparable and therefore combinable.

Bank innovation

Measuring innovation is difficult primarily because it is not readily identifiable or observable. It is common to use the number of patents and their citation numbers to measure innovation. This is more uncommon for the financial industry, with exceptions for the United States. In 1998 the US legal system approved patents on business methods in the ‘State Street’ Case. However, as of 2000 they were mostly granted to companies rather than universities and the number of granted patents skyrocketed (Lerner, 2002). The availability of patents should have a positive effect on financial innovation (Cornaggia, Mao, Tian, & Wolfe, 2015). For other parts of the world using patent data is not an option.

(28)

Since this thesis is a global analysis an alternative was needed. It is impossible to measure innovation directly, but possible to determine the result. This is measurement is possible since the Schumpeterian definition of innovation is used, which includes market success in a broad way. Therefore, if a country has more evolved and functioning financial

sector, more innovation has taken place. Knowing that regulatory systems are relatively

stable over time, it is possible to measure which systems function better than others.

Although Demirgüç-Kunt & Klapper (2012) create a lot of statistics on financial inclusion, there is not one particular variable that is suitable to use as a dependent variable. I’ll create a variable that includes all the relevant measurements in a consistent measure. Since the data takes different forms (e.g. percentages, rankings, point scales) it is impossible to simply average or sum them. For the analysis a variable ‘Innovation’ is required that takes higher values for a country with a more advances financial system and lower for less advanced countries. The nine chosen statistics (see appendix) will be ranked in 10 deciles with, regulating in a rank 1-10. Rank 10 will always be the 10% most advanced, whereas 1 will be 10% least advanced countries.

It is now possible to sum the results and create an index for advancement per country that is suitable for regression. Demirgüç-Kunt & Klapper (2012) state that there is strong correlation with GDP on many accounts, so it is needed to control for that by including this into the regression. Hence the presence of the GNI Indicator. Still, especially among lower income economies there are large differences among countries with the same GNI per capita.

(29)

The selection of data is based on other literature (Chaia et al., 2013) and extended due to differences in focus of the studies. The selected data shows the general level of inclusiveness for:

• Retail banking

o Consumer accounts

o Consumer deposits (wages paid in checking account) o Availability of ATM’s

o Availability of point of sale equipment o Bank branches

o Mobile accounts • General

o Strength of legal rights o Credit to Private sector

o Availability of credit information.

This data is able to give a general overview of the inclusiveness of the various countries.

Bank supervision.

Just like bank innovation, bank regulation is also very hard to measure. Using the World Bank Regulation Survey does at least give us consistent data on what regulations are present worldwide. Not all countries have filled in the survey, but in 2011 (the survey used by this thesis) 142 countries responded. As input the variable will not use the raw data provided by the World Bank but the dataset of the analysis by Barth et al. (2013). Barth et al. (2013) uses 10 different categories of variables, which will also be used in this thesis.

1. Bank Activity Regulatory 2. Financial Conglomerate 3. Competition Regulatory

(30)

4. Capital Regulatory

5. Official Supervisory Action 6. Official Supervisory Structural 7. Private Monitoring

8. Deposit Insurance Scheme 9. Market Structure Indicators 10. External Governance Variables

Since some of the variables overlap substantially, the variables were contracted by using the description listed in the dataset, not the ten that Barth et al. (2013) constructed.

Although regulatory structures don’t change regularly of suddenly this still occurs in some countries. After the financial crisis some countries have become stricter, and some became looser in their regulatory practices. An extra dummy to indicate this has been introduced in the cross section (Barth et al., 2013). The World Bank also distinguishes between several types of business conduct and prudential supervision. Both have been added as dummies to the regression to control for the type of regulation. Since financial innovation is closely linked to income per capita (see figure 1) this is added as a control variable. This helps control possible external validity problems.

Data cleaning

The data cleaning process is straightforward. Starting off with the initial dataset provided by the analysis of Barth et al. (2013) of the World Bank Regulation Survey. Countries that had not or just partially (<50%) filled out the survey were deleted. After combining the datasets all countries that missed the dependent variable were also deleted. This resulted in a sample of 120 countries with data from 2011, from all habitable continents. For the large regression, missing values were left blank, whereas for the partial regressions they were filled with the average values of the other countries inside the GNI bracket. This enables us to continue with

(31)

the smaller regressions. It is necessary since missing values have larger influence on smaller samples.

Table 1 is showing the process of data cleaning from the variables. Starting with the Regulation Dataset with 180 countries, where countries that did not fill out the 2011 questionnaire were deleted. Countries that did fill out the questionnaire but were not available in the Financial Development database were subsequently deleted. Resulting in 120 countries in the final dataset.

Action Countries left in sample

Start countries: 180

No answers 143

Matching with Dependent

variable 120

Table 2 shows the continents that the remaining 120 countries of the sample belong to. Contrary to other publications in this field, it consists of 68% non-western countries, giving insight in these continents. Additionally, this is where financial innovation is most important. Percentage Sum ≠ 100% due to

rounding differences. Continent Count % Africa 29 24% Asia 20 17% Oceania 2 2% Europe 36 30% North-America 2 2% Middle-East 11 9% South-America 20 17% Sum 120 101%

(32)

Creating regulator types

The regulatory system per country is assessed by its two main characteristics: power and stringency. There is no current format for qualifying financial regulatory systems worldwide. The two main determinants of a financial regulatory system are the ability to have an influence (power) and the extend of the influence of the regulator (stringency). There are four questions in the survey directly indicating power in the questionnaire, where 14 relate to stringency and restrictiveness. Although Barth et al. (2013) quantifies values for each question it does not rank regulatory systems. This is done by calculating the value of each quantified question in divided by the highest obtained value for that question. This gives a relative score to each of the questions. Then the values corresponding to the type of questions (e.g. power) are all averaged to reflect a definite relative number to other regulatory systems. Since the average answer rate per question is around 75%, it is impossible to sum the values in order to reach a comparable number. The resulting two by two matrix plots divides regulatory systems into four groups.

This is the first regulatory system qualification made on a worldwide basis. Several authors (Barth 2012, Jackson 2007) have classified groups with similar regulations, but not different regulations with the same extend of strictness or power. This will enable analysis on the superiority of regulatory systems and give first insight to the division of regulatory practices worldwide.

(33)

Figure 4 Shows the regulatory Strictness and Ability Matrix. The horizontal axis shows the score on Regulatory Strictness as calculated in the Independent Variable. It indicates the strictness of regulatory policy. The vertical axis is created by using data from the independent variable of Supervisory Power, which indicates to what extend market supervisors are able to intervene in the financial Sector. The resulting Matrix shows the strictness of regulation and the ability for regulators to act on those strict regulations. Data: World Bank Regulatory Survey

Regulatory Strictness à

Supe

rvi

sor

y

pow

er

à

Not Strict /

Able

Strict /

Able

Not Strict /

Not Able

Strict / Not

Able

Creating Variables Independent

The eight variables are based upon the quantification of questions done by Barth et al. (2013). The variables are created by analyzing the quantification explanation of questions. These are grouped by what the quantification is relating to. Most quantifications state for example “Higher values indicate more restrictive”. This is used to group questions and thus data that indicates similar questions. Then all variables were modified to be positively related. For example: If a question was quantified with ‘higher number indicates a less strict regulator’ it was transformed into ‘higher number indicates a stricter regulator’. This resulted in transforming the data connected on a 0-8 scale an ‘8’ value was transformed to an ‘8’ and vice versa.

(34)

Values were standardized by using: 𝑆𝑡𝑎𝑛𝑑𝑎𝑟𝑑𝑖𝑧𝑒𝑑 𝑉𝑎𝑙𝑢𝑒 = 01234 563789:

;1< 01234 6= 019>1?24where

the outcome is ≥ 0 ≤ 1. This results in comparable values. After grouping of the questions,

the average of these standardized values was calculated and used as the variable value in the regression. Average is used instead of the sum, so that variables would be consistent when one value was missing.

Restrictiveness

Restrictiveness denotes higher values for more restrictive regulations on banking. It consists of questions I.I – I.IV and II.I – II.IV. These questions relate to the type of activities banks are allowed to offer to their clients. In research based on 30 OECD countries it was found that more restrictive regulation is counterproductive on innovation (Bartelsman, Gautier & Wind., 2011). It also shows the reluctance of firms to adapt to new technology.

Stringency

This variable shows higher values for stricter regulations on allowing foreign bank ownership, entry requirements, loan classification and capital regulations. It consists of questions IV.I – IV.III, V.VII and V.VIII Increased stringency is most likely to result in incremental rather than radical innovation (Blind, 2012). In our dependent variable, this would lead to lower values of bank innovation. In advanced economies it could be a consequence of financial innovation.

Applications

The variable Applications has higher values for a larger percentage of foreign and domestic denied applications for bank licenses, either by acquisition or start up. It consists of questions

(35)

III.III – III.VI. No research has been done on this relationship. It is however a good indicator of the ease of entrance to the financial system in this market.

Power

Power has higher values for regulators with more possibilities to take action when regulations are breached, the ability to impose fines, declare insolvency or revoke bank licenses. It consists of questions: V.I, V.III, V.IV and VIII.I. There is no research on this relationship. More power could either constrain the freedom of movement for innovation, or promote trust in the banking system. This could also differ among developed of undeveloped markets.

Control

The variable Control shows higher values for markets with more mandated external controls. This includes mandatory audits, ratings by international rating agency’s and the availability of a deposit insurance scheme. The variable consists of questions: VII.I – VII.VI. Higher control imposes higher compliance costs for banks, but will also greatly improve the trust in the banking system.

Audit

Audit has higher values for stricter Audit regulations for banks and better credit monitoring. Includes the transparency of financial statements and compliance with IFRS standards. It consists of questions: X.I – X.V.

Structure

The variable Structure has higher values for more concentrated bank markets. It includes the % of deposits in the largest 5 banks and the % of assets concentrated in the largest 5 banks. It

(36)

is a measure for the amount of players and therefore competition in markets. It is made up of questions IX.I – IX.IV.

Discretion

Discretion shows higher values for less regulatory discretion. Indicates whether or not the supervisor is capable and allowed to rule over situations as bankruptcy, forbearance of rules and exclusions from regulations without court or political approval. It consists of questions V.V and V.VI

GNI per Capita

GNI per capita has higher values for countries with a higher Gross National Income per capita. The 2011 data comes from the International Monetary fund.

Table 3 Descriptive statistics for the independent variables and control variable as used in the regression. The different maximum value for Stringency is due to a country that was not used in the regression but is present in the final datasheet. The Jarque Bera test shows that four out of nine of the variables are normally distributed. The control variable however is not close to normally distributed, and suffers from high Kurtosis levels, indicating heavy tails in the distribution.

Structure Stringency Restrictive Power Discretion Control Audit

Appli-‐ cations

GNI per Capita (Control) Mean 0.52 0.66 0.64 0.57 0.36 0.65 0.81 0.15 16477 Median 0.50 0.66 0.66 0.61 0.33 0.64 0.87 0.00 7305 Maximum 0.92 0.88 1.00 1.00 1.00 1.00 1.00 1.00 102700 Minimum 0.25 0.32 0.32 0.00 0.00 0.25 0.36 0.00 260 Std. Dev. 0.13 0.10 0.15 0.27 0.18 0.15 0.17 0.24 20911 Skewness 0.24 -‐0.34 0.01 -‐0.37 0.64 -‐0.19 -‐0.71 1.72 1.88 Kurtosis 2.96 3.29 2.35 2.32 3.73 2.76 2.34 5.15 6.40 Jarque-‐Bera 0.99 2.80 2.09 4.91 10.65 1.02 12.27 66.35 128.52 Probability 0.61 0.25 0.35 0.09 0.00 0.60 0.00 0.00 0.00 Sum Sq. Dev. 1.70 1.29 2.55 8.41 3.82 2.63 3.36 5.33 52034112879 Observations 102 121 119 117 119 121 120 97 120

(37)

Dependent

The dependent variable shows the the level of financial advancement of a country in 2011. This year corresponds with the year of the date used in the independent variables. Nine indicators of financial development are combined to create an index number per country. Values were standardized by using: 𝑆𝑡𝑎𝑛𝑑𝑎𝑟𝑑𝑖𝑧𝑒𝑑 𝑉𝑎𝑙𝑢𝑒 = _{;1<
01234
6=
019>1?24}01234 563789: where the

outcome is ≥ 0 ≤ 1 . This results in comparable values. After grouping of the variables the

average of these standardized values was calculated and used as the variable value in the regression. Average is used instead of the sum, so that variables would be consistent when one value was missing.

Table 4 shows the descriptive statistics for the dependent variable as used in the regression. Since Kurtosis levels are close to 3 and Skewness levels are very close to 0, the dependent variable is closed to normally distributed.

Descriptive statistics Financial Development 2011

Mean 0.34 Kurtosis 2.46

Median 0.35 Jarque-‐Bera 1.50

Maximum 0.67 Probability 0.47

Minimum 0.02 Sum 40.85

Std. Dev. 0.15 Sum Sq. Dev. 2.75

Skewness 0.04 Observations 121

Regression Equation

1 𝐹𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑆𝑒𝑐𝑡𝑜𝑟 𝐼𝑛𝑛𝑜𝑣𝑎𝑡𝑖𝑜𝑛 = 𝛽_J+ R𝑒𝑠𝑡𝑟𝑖𝑐𝑡𝑖𝑣𝑛𝑒𝑠𝑠_>8𝛽_N+ 𝑆𝑡𝑟𝑖𝑛𝑔𝑒𝑛𝑐𝑦_>8𝛽_Q+ 𝐴𝑝𝑝𝑙𝑖𝑐𝑎𝑡𝑖𝑜𝑛𝑠_>8𝛽_T+ 𝑃𝑜𝑤𝑒𝑟_>8𝛽_W+ 𝐶𝑜𝑛𝑡𝑟𝑜𝑙_>8𝛽_Y+ 𝐴𝑢𝑑𝑖𝑡_>8𝛽_Z+ 𝑆𝑡𝑟𝑢𝑐𝑡𝑢𝑟𝑒_>8𝛽_[+ 𝐷𝑖𝑠𝑐𝑟𝑒𝑡𝑖𝑜𝑛>8𝛽]+ 𝐺𝑁𝐼 𝑝𝑒𝑟 𝐶𝑎𝑝𝑖𝑡𝑎>8𝛽`+ 𝑢>8

(38)

The model has been estimated and checked for heteroscedasticity and endogeneity problems. The large sample model and several of the smaller sample models showed signs of heteroscedasticity, which was corrected for using White Standard Errors. The large model holds all available countries. However, since there are differences among various country groups, it is necessary to test for specific relationships among these countries. Subsets have been created for these groups. Specific groups were created for each continent and three GNI classes: low-, middle and high income. The per country qualifications from the regulatory matrix was also tested separately. Since three countries are not rated in the matrix the total number of observations in 117.

Geographical groups have been tested alone or with comparable income groups. This is due to the fact that smaller samples where non-parametric variables are present are not suited for statistical analysis. Groups will be 29 or higher to be suitable for statistical analysis. Tested groups are:

• Entire Sample (120) • Africa (29)

• Europe & North America (38) • Asia & South-America (40)

• Lower third income economies (40) • Middle third income economies (40) • Highest third income economies (40)

• Not Strict / Unable Regulators (29) • Not Strict / Able Regulators (32) • Strict / Unable Regulators (29) • Strict / Able Regulators (37)

(39)

Empirical Results

The empirical results show strong relations between Market Structure, Regulatory Strictness and GNI per capita. The results differ among continents, income classes and regulatory types. Those regulatory types come from the Regulatory Matrix, that consists of Power and Stringency values.

Power and Stringency Matrix

The matrix (shown in Figure 6) shows a widely distributed area of qualifications for countries. The wide disparity shows the ability of the matrix to highlight returns among different regulatory systems. There is no apparent relation to the location of the regulator and its position in the matrix. When aggregating the values per continent, differences appear primarily in Power (see figure 5). This ability to intervene is much larger in North-American regulators and noticeably smaller in Middle-Eastern regulators. European regulators show lower values on Stringency compared to the rest of the world, whereas African regulators score stricter values. The matrix also shows tendency for regulators to be quite powerful. Also noteworthy is that all middle-eastern regulators except for Kuwait score in the highest values for stringency.

(40)

Figure 5 Shows the Power and Stringency values averaged per continent. Note that in Europe, South-America and Oceania the values are evenly distributed. Where in Africa, Asia and the Middle east, Stringency has much higher values than power. This is reversed in North-America. Vertical axis values are not present due to the type of data modification, which results in comparable values.

(41)

Figure 6 Shows the regulatory Strictness and Ability Matrix. The horizontal axis shows the score on Regulatory Strictness as calculated in the Independent Variable. It indicates the strictness of regulatory policy. The vertical axis is created by using data from the independent variable of Supervisory Power, which indicates to what extend market supervisors are able to intervene in the financial Sector. The resulting Matrix shows the strictness of regulation and the ability for regulators to act on those strict regulations. Colors indicate country’s continent. The full list is present in the appendix. Data: World Bank Regulatory Survey

(42)

Regression Results3

Table 5 shows the overview of all regressions for various sample groups. It shows results for regression (1) and also shows the coefficient sign when variables show a significant relation with the adaptation to financial innovation.

3_{For this chapter, all mentioned but not presented regression result tables are shown in the}

appendix.

Sample type Sample

Variable significant

at α= 0.01 or α=0.05 Relation

Entire Sample Restrictiveness -

Structure -

GNI per Capita +

Continents Africa Applications +

Restrictiveness +

Structure -

GNI per Capita +

Europe & N-America None

Asia & S-America None

Income per capita Lower Third Income None

Middle Third None

Highest Third Structure -

Regulatory Matrix Not Able / Not Strict GNI per Capita +

Able / Not Strict GNI per Capita +

Strict / Not Able GNI per Capita +

Strict / Able Audit -

Control +

Stringency -

Structure -

GNI per Capita +

All Strict Stringency +

Structure -

All Able Stringency -

Structure -

Adaptation to Financial Innovation

Lower third None

Middle Third None

(43)

Entire sample regression.

The regression results for the entire sample show significant relationships for three variables, including the control variable. Restrictiveness show the most extreme negative coefficient of the regression. It shows that countries where banks are allowed to offer less types of services to their customers, show lower ratings for the adaptation to financial innovations. This would mean that looser regulation on the type of activities that banks are allowed to offer would enhance the adaptation to financial innovation.

The negative coefficient for Structure indicates that adaptation suffers when there is a more concentrated market. It takes higher values when there are more assets and clients at the countries top five banks, indicating les competition in the market. GNI per capita is significant, as expected. It is not possible to state anything on the relationships between the other variables and adaptation. Since they are not statistically significant.

Table 6 Regression Full Sample results of the regression for all values in the dataset. Three variables including the control variable show significant results.

Variable Coefficient Significance Std. Error

Applications 0.02 0.0543 Audit 0.02 0.0763 Control 0.05 0.0964 Discretion -0.02 0.0648 Power 0.03 0.0587 Restrictiveness -0.23 * 0.0834 Stringency -0.16 0.1524 Structure -0.22 ** 0.1093

GNI per Capita 0.00 * 0.0000

Observations 120 Log likelihood 66.14

Adjusted R-squared 0.33 F-statistic 5.36

S.E. of regression 0.11 Prob(F-statistic) 0.00

Sum squared resid 0.93 White Std Errors Yes

* significant at the 1% level, ** significant at the 5% level *** significant at the 10% level

The influence of regulatory and supervisory practices on worldwide adaptation to financial innovation

Master Thesis